3.2.1 Evidence from the Registries

The American National Registry of Exonerations records that mistaken eyewitness identification played a role in 987 of 3,649, or 27 per cent of exonerations as of the end of 2024. This is a much lower percentage than found in DNA exonerations that are typically stranger rapes and murders (Garrett, Reference Garrett2011a: 48).

The British registry as of the end of 2024 records 26 of 494 wrongful convictions related to eyewitness identification and another 14 related to eyewitnesses in general and another 22 related to other eyewitnesses. All of these identification errors amount to only 12.5 per cent of remedied wrongful convictions.

The Canadian registry records twenty-five of eighty-nine wrongful convictions related to mistaken eyewitness identification. This amounts to 28 per cent of remedied wrongful convictions, a figure remarkably similar to that obtained from the much larger American registry.

The European registry records, as of the end of 2024, 34 of 136 or 25 per cent of its cases as flawed eyewitness identification (procedure). This percentage is also quite close to that in the American and Canadian registries.

3.2.2 Legislative Regulation

In part in response to well-publicized wrongful convictions, the Police and Criminal Evidence Act, 1984 (PACE), was enacted in England. It is accompanied by detailed codes of practice that are directed at the police and periodically updated. Code D applies to identification procedures. It provides for double-blind procedures, sequential showing of at least twelve photos, and required statements to the eyewitness that the suspect may not be present. It also provides for the attendance of the suspect’s lawyer at the line-up and disclosure of any photos or sketches shown to the witness before the identification procedures. At the same time, it allows individual showups where a witness is essentially asked to identify a single person and group identifications, where they are asked to identify a person in an informal group if the previous procedures are not practicable. This reflects an unwillingness to forgo these forms of identification, which police frequently use, and the Code allows for considerable police discretion in deciding when these show-ups should be used. At the same time, Code D still regulates showups with an appendix providing:

Even if the Code is breached, exclusion of an identification is far from automatic under PACE. Courts have admitted some identifications taken in breach of the code in cases where they stressed the strength of the identification evidence (R. v. Forbes [2001] UKHL 40; R. v. Lariba [2015] EWCA Crim 478; DPP v. Jobling [2016] EWHC 2707 (Admin)). The unwillingness to prohibit the use of identifications obtained in violation of the PACE procedure reflects an unwillingness to sacrifice identifications and possible convictions obtained through less than optimal procedures.

Despite numerous calls for legislative reform, Canada still has not enacted any legislation to govern identification (or interrogation) practices. The federal Parliament has national jurisdiction over criminal procedure, but the inquiries calling for national laws on identification procedures have been held by the provinces which have primary jurisdiction over policing (Campbell, Reference Campbell2018: 37–49; Cory, Reference Cory2001). This, as well as the lack of political popularity of wrongful conviction reform legislation and legislative abdication of the regulation of police conduct to courts enforcing a constitutional bill of rights, explains Canada’s lack of legislation (Roach, Reference Roach2022b: ch 4; Roach, Reference Roach2023a: ch 10).

There are a variety of different approaches to legislative regulation in American states, and this supports the idea that federalism can promote democratic experimentation. Many states, including California, North Carolina, Georgia and New York, have laws which mandate similar double-blind and sequential procedures as also required by PACE. These were generally enacted much later than the 1984 English legislation with North Carolina enacting its law in 2008, New York enacting its law in 2017 and California only enacting its law in 2020 after earlier attempts were vetoed by Governor Schwarzenegger (Findley, Reference Findley2016: 421).

One of the earliest American laws was 2006 legislation enacted in Wisconsin that simply required police forces to develop eyewitness identification policies and encouraged them to follow a model policy developed by the state Attorney General. Keith Findley (Reference Findley2016: 435) has examined the response of Wisconsin police departments and found that it lacked consistency and adequate oversight.

Many of the American laws regulating identification procedures were obtained through lobbying by Innocence Projects, who often worked with the police to avoid possible resistance and undermining of the regulation. They also sometimes assisted with police training and engaged with litigation aimed at problematic identification practices (Brown and Saloom, Reference Brown and Saloom2013). This points to the need not simply to persuade the legislature, but also the executive and the judiciary. Keith Findley (Reference Findley2016) has persuasively argued that such hybrid approaches have proven superior to approaches that rely solely on the command and control of legislation or voluntary compliance.

The Codes of Criminal Procedure used in various inquisitorial systems can regulate identification procedures. The Argentinian Code of Criminal Procedure requires that the witness describe the suspect before a line-up and that the line-up not be suggestive. A failure to comply with these requirements will generally not result in the exclusion of the identification. Defence counsel is required at Mexican line-ups, but a failure to have counsel present will not result in the exclusion of the identification (Bradley, Reference Bradley2007: 26, 362–363). The true impact of rules requires appreciation of the remedies or lack of remedies for their violation (Dicey, Reference Dicey1885; Levinson, Reference Levinson1999; Roach, Reference Roach2021).

There are no formal identification rules in the German and French Codes (Bradley. Reference Bradley2007: 214, 255), but the Italian Code regulates three different procedures: (1) a summary stop or show-up where a person is required to identify themselves and the police are able to verify the identification, (2) a pre-trial parade conducted by the prosecutor where defence counsel does not have to be present and (3) a judicial recognition where the judge questions the witness about any prior recognition with defence counsel present and at least two foils are used (Reference Bradley2007: 322–324). Attempts to regulate show-ups are probably better than legislative reforms that ignore the subject altogether.

Bosnia’s Code of Criminal Procedure disallows show-ups and requires only live or photo line-ups with at least five foils subject and the right to defence counsel. Unlike in many countries described earlier, these rules are enforced by a strict exclusionary rule. Carrie Leonetti (Reference Leonetti2014) has concluded that this approach is supported by psychological research showing that multiple-person line-ups are more reliable and less suggestible than one-person line-ups or show-ups. She quotes a Bosnian police officer who stated that a line-up is a better procedure, “especially if you want to find the truth” (Reference Leonetti2014: 475). This ban on show-ups is exceptional.

Another problem with eyewitness identification is the frequent use of in-court identification of the accused. This also reflects a larger problem, as witness often becomes more certain of their identification of the accused over time, especially when they have made multiple identifications. In 2017, Texas enacted a law that provided that no witness would be able to give an in-court identification of the accused as the perpetrator of the crime unless other and often more reliable evidence of prior identifications (or the lack thereof) were introduced in court (Texas Code of Criminal Procedure Article 38.2). In Chapter 6 on the American experience, it will be suggested that both Texas and North Carolina have been at the forefront in legislative wrongful conviction reforms. Both states have had frequent exonerations and retain the death penalty. These factors may help explain such bipartisan reforms.

3.2.3 Executive Regulation

In 2001, New Jersey became the first state to adopt detailed guidelines issued by the state Attorney General based on best practices with respect to eyewitness identification, including the use of double-blind procedures. At the same time, the guidelines provided that failure to follow them did not affect the admissibility of an identification in court. New Jersey is interesting because the regulation of identification procedures has come from the executive and the judiciary, not the legislature.

Garrett (Reference Garrett2011a: 82) reported that many American police departments have adopted double-blind procedures or a less labour-intensive procedure of allowing eyewitnesses to shuffle photos in a sequential line-up. A Canadian report indicated that 85 per cent of responding police agencies offered some training on eyewitness identification. Unfortunately, it provided no detail about actual implementation (FPT, 2018: 32). It recommended the use of double-blind procedures “if possible” (2018: 42). As in Wisconsin, there is a danger that allowing the police to self-regulate will not be effective, especially if there is no oversight. At the same time, some degree of police buy-in and acceptance of regulation seems essential (Findley, Reference Findley2016). Another possible area of executive reforms is better training of police dispatchers to secure the most accurate possible description of suspects (Brown and Neufeld, Reference Brown and Neufeld2021: 254).

In 2001, the Taiwan Ministry of Justice and National Police Agency introduced guidelines for double-blind photo line-ups and restricting show-ups which have been described as “very close to the best standards proposed by social science scholars.” Yet in at least one remedied wrongful conviction in 2017, the police did not follow these procedures and used a one-person photo line-up. Taiwan’s Supreme Court has ruled that even an exclusionary rule based on balancing rights against public interest did not apply because the administrative guidelines do not have the force of law (Chien, Reference Chien2022). Executive regulation can be helpful but may also require judicial oversight to be effective (Findley, Reference Findley2016).

3.2.4 Judicial Regulation

The United States Supreme Court has attempted to regulate suggestive show-ups, but in almost every major case has refused to exclude identifications obtained through suggestive procedures. In 1967, the Court admitted evidence from a suggestive show-up where the Black accused was the only Black person in the room. It ruled that even suggestive identifications could be admitted if the court determines that the identification was sufficiently reliable (Stovall v. Denno 388 U.S. 293 (1967)). In 1968, the Court allowed the use of a single photo line-up with five witnesses a day after a bank robbery (Simons v. U.S. 390 U.S. 377 (1968)). In 1972, the Court held that a rape victim’s certainty about the suspect’s identity at a show-up was a sufficient indicia of reliability (Neil v. Biggers 409 U.S. 188 (1972)). Subsequent DNA exonerations, however, indicate that certainty does not always equate to reliability.

Independent courts may in theory be willing to exclude evidence of dubious reliability, but the United States Supreme Court had been cautious about using the exclusionary remedy with respect to eyewitness identifications. It will not exclude the result of a suggestive identification practice if there are other indicia of reliability, including the witness’s confidence, accuracy of description and original opportunity to view (Manson v. Braithwaite 432 U.S. 98 (1977)). It has not updated these indicia of reliability to take into account the vast psychological literature or DNA exonerations. Judicial regulation may often lag behind the evolution of the underlying science.

In 2011, the majority of a conservative Court ignored much evidence about mistaken eyewitness identification being an immediate cause of wrongful convictions in DNA exonerations and refused to exclude a suggestive identification. Justice Ginsburg, for the Court, rejected a due process right that evidence be reliable. She stressed the need to defer both to state processes and the jury in making determinations of reliability. The Court accepted a witness’s identification of a Black man even though the witness could not subsequently identify the suspect in a photo line-up. Because the police had not arranged the suggestive show-up, there was no state action as required to enforce the Bill of Rights or need to deter police misconduct (Perry v. New Hampshire 565 U.S. 228 (2012)). Justice Sotomayor, in her lone dissent, cited the role of mistaken eyewitness identification in 76 per cent of the first DNA exonerations and vast amounts of scientific research on the frailties of eyewitness identification.

The right to counsel may apply to physical line-ups (US v. Wade 388 U.S. 218 (1967)), but not to more frequently used photo line-ups (Kirby v. Illinois 406 U.S. 682 (1972)) or in cases involving witnesses (United States v. Ash 413 U.S. 300 (1973)). The right to counsel is at best an indirect way to ensure the reliability of identifications. The large American jurisprudence on this issue suggests that due process-style judicial regulation can fail to focus on reliability issues (Findley, Reference Findley2009a).

Federalism in the United States allows judicial as well as legislative experimentation. The New Jersey courts have required judicial instructions on the frailties of cross-racial identifications (State v. Cromedy 727 A.2d 457 (N.J., 1999)). A decade after the Attorney General’s guidelines, they updated them and required pre-trial judicial screening of identifications and updated warnings to juries to reflect current research (State v. Henderson, 27 A. 3d 872 (N.J., 2011)). At the same time, it also rejected requests that any departures from guidelines result in the automatic exclusion of the identification as too drastic given competing social interests. It also contemplated both the admission of expert evidence and tailored instructions to the jury as less drastic alternatives to the exclusion of an identification. The judgment only had a prospective effect in order not to disrupt the finality of many convictions. Judicial regulation can be more effective than voluntary measures, but it will also not be oblivious to competing social interests.

After receiving briefs from the Innocence Project and academics researching eyewitness identification, the Oregon Supreme Court issued a similarly comprehensive opinion as the New Jersey court (State v. Lawson (291 P. 3d. 673 2012)). The Court approved of the acceptance of expert testimony about over a century of research on the frailties of eyewitness identification. Jennifer Mnookin (Reference Mnookin2015) has recommended that one possible way to mitigate the costs of such expert evidence is to use pre-recorded testimony. In general, identification reforms have not caught up to our digital age.

The Wisconsin courts in 2005 ruled that because of their inherently suggestive nature, show-ups should only be used in exigent circumstances when line-ups are not possible (State v. Dubose 699 N.W. 2d 582). Professor Findley (Reference Findley2016: 446–447) found that in response, most police forces developed policies on the use of show-ups, but that about a quarter of those policies did not comply with the Court’s restrictions on their use. He cited this as a need for hybrid strategies that combine both judicial and executive reforms.

Canadian courts have continued to reject expert testimony on the frailties of eyewitness identification and accepted identifications even when the police showed the eyewitness a photo of the Indigenous suspect on social media (R. v. Delorme 2017 SKCA 3). In a 2007 decision, however, the Supreme Court excluded eyewitness identification enhanced by hypnosis as insufficiently reliable to be used in a criminal trial. In doing so, the Court specifically cited concerns about wrongful convictions (R. v. Trochym 2007 SCC 6). Garrett (Reference Garrett2011a: 61) found that in five of the first 250 DNA exonerations, hypnosis had been used to enhance a witness’s eyewitness identification despite studies showing that it could make the witness more open to suggestions. Canadian courts have also refused to prohibit in-dock identifications while also requiring judges to give judges strong warnings that such suggestive identification procedures are not reliable (R. v. Hibbert 2002 SCC 39).

The British courts rely on warnings and guidance to juries about the frailties of eyewitness identification. The failure to provide such warnings can lead to a conviction being quashed as unsafe (R v. Turnbull [1977] QB 224; Farquharson v. The Queen (1993) 98 Cr App R 398). At the same time, judicial warnings focus on how long the witness observed the suspect and stopped short of a 1976 public inquiry conducted by Lord Devlin, which recommended that it would be unsafe to convict on uncorroborated eyewitness identifications. There is a need to ensure that all forms of regulation are regularly updated to take into account new research and new technology. English judges also have a discretion to exclude identifications that were obtained by the police in breach of Code D of PACE. The discretion will depend on the judge’s determination of the seriousness of the breach and its effect on the fairness of the trial.

The Judicial Committee of the Privy Council has placed restrictions on the frequent use of in-dock identifications in cases where they retain jurisdiction in the British Caribbean. It has not prohibited such a suggestive procedure but has stressed that the trial judges have a discretion to disallow it and that if they do allow it, they should warn juries about the ease of identifying the accused in court (Aurelio Pop v. The Queen [2003] UKPC 40; Holland v. H. M. Advocate [2005] UKPC D1; Pipersburgh and Another v. The Queen [2008] UKPC 11; Tido v. The Queen [2012] 1 WLR 115; Neilly v. The Queen [2012] UKPC 12; Lawrence v. The Queen [2014] UKPC 2). In one death penalty case from Trinidad and Tobago, Lady Hale dissented on the basis that the jury should have been warned about the police’s failure to conduct a less suggestive identification procedure before an in-dock identification of the accused in court was allowed (John v. Trinidad and Tobago 2009 UKPC 12). Unfortunately, case-by-case judicial regulation may not necessarily encourage the police to be trained to use less suggestive identification procedures.

3.2.5 Summary

Despite the fact that it was enacted in 1984 before the first DNA exonerations, the English PACE remains a gold standard in terms of regulating police conduct that is the immediate cause of wrongful convictions (Bradley, Reference Bradley1993). As discussed earlier, identification errors constitute a cause in only 12.5 per cent of remedied wrongful convictions in the UK registry but 25 per cent or more in registries in Canada, Europe and the United States.

PACE combines both comprehensive regulation of identification practices and a Code of Practice, which provides specific guidance to the police that can be easily updated. Code D has also been updated to take account of increased use of video evidence, something not done in other jurisdictions. It also regulates police release of images of suspects. It enables the police to use show-ups but also attempts to regulate them in a way that, if breached, can result in either the exclusion of the evidence or informing the jury about the breach of the procedure. In this way, PACE combines legislative, judicial and executive regulation, whereas much of the other regulations only involve one branch of government.

There is a strong scientific basis for best practices such as the use of sequential and double-blind photo or video identifications (Findley, Reference Findley2016). A good number of jurisdictions have implemented these reforms. At the same time, none seem to mandate automatic exclusion of identifications obtained in violation of these best practices. Bosnia seems to be the only country that categorically prohibits all suggestive show-ups (Leonetti, Reference Leonetti2014). The performance of the United States Supreme Court on suggestive show-ups has been particularly disappointing. Consistent with the idea of American extra-legalism (Roach, Reference Roach2011: ch 4), it has produced a complex jurisprudence, but one that almost always admits identifications as evidence even after the state has used suggestive identification procedures. This underlines that competing social interests will not be ignored in wrongful conviction reforms.

The emphasis in most countries has been on the regulation of formal identification procedures. This can provide an illusion of reform if the police rely on less regulated show-ups or identification of video images. The bottom line is that there is little ground to believe that eyewitness identification regulation by any branch of government will prevent mistaken identifications and wrongful convictions based on such errors.

3.3.1 Evidence from the Registries

The American registry as of the end of 2024 detailed 503 cases where a co-accused implicated an exoneree and 256 cases where a jailhouse informant implicated the exoneree out of 3,649 wrongful convictions. Moreover, it listed 2,344 out of 3,649 remedied wrongful convictions where perjury or false accusations were made by someone other than the accused. This latter figure reveals the extent of the problem by indicating that witnesses who lie are a factor in over two-thirds of remedied wrongful convictions. Punitive American penal culture encourages snitching.

The British registry as of the end of 2024 records 42 out of 494 remedied wrongful convictions involving testimony from an informant or co-accused. This suggests that informant or co-accused testimony was a cause of only 8.5 per cent of remedied wrongful convictions compared to 20.8 per cent of all remedied wrongful convictions in the United States.

As with the American registry, the British registry reveals that informants or co-accused are only a minority of the inaccurate witnesses who contribute to wrongful convictions. Witness evidence from non-complainants was a factor in 122 of 494 wrongful convictions, and witness evidence from a complainant was a factor in 58 cases. These are high percentages but lower than in the United States.

The Canadian Registry of Wrongful Conviction has thirteen cases that used jailhouse informants and twelve cases (with only one case overlapping) that used incentivized informants. In addition, another seventeen cases involved false accusations made by a non-police witness out of eighty-nine remedied wrongful convictions as of the end of 2024. The Canadian experience is noteworthy because almost 15 per cent of remedied wrongful convictions involved jailhouse informants, more than double the 7 per cent of remedied wrongful convictions in the United States. Three of Canada’s seven public inquiries into wrongful conviction made recommendations about banning or limiting the use of jailhouse informants.

The European Registry does not list jailhouse informants as a separate immediate cause, though perjury and false accusations make up 43 of 136 remedied wrongful convictions as of the end of 2024. All of the registries reveal lying witnesses as a recurring and intractable cause of wrongful convictions.

3.3.2 Legislative Regulation

Although one inquiry into a Canadian wrongful conviction that involved three unreliable jailhouse informants (and eleven more who had also volunteered their services) recommended that a total ban should be enacted on the use of jailhouse informants because they were a “cancerous corruption of the administration of justice” (Cory, Reference Cory2001), the Canadian Parliament has not enacted or considered enacting such a ban or regulating the use of such witnesses. Another Canadian wrongful conviction inquiry recommended that testimony from jailhouse informants be presumptively inadmissible (Kaufman, Reference Kaufman1998: recommendation 59), but that recommendation has also not been adopted by either legislatures or courts. Following a pattern seen with respect to eyewitness identification, recommendations made by provincial public inquiries have not resulted in federal criminal procedure laws (Roach, Reference Roach2023a: ch 10).

Jailhouse informants are not categorically banned under American legislation, though bills to ban all “criminal informants” were considered but not passed in New York and Texas. California prohibits the use of informants under thirteen years of age and requires judicial approval of those under eighteen years of age. North Dakota bars the use of informants under sixteen years of age, and laws in Florida and Minnesota direct police to consider the reliability of informants who may have addictions. These laws were enacted not so much in response to concerns about wrongful convictions but well-publicized cases where informants were murdered or killed themselves (Natapoff, Reference Natapoff2022: 186–187, 190–191). This raises questions about whether remedied wrongful convictions create enough publicity and public concern to motivate legislative reform.

Illinois enacted laws requiring judges to conduct pre-trial hearings about the reliability of statements from jailhouse informants. It also provided enhanced disclosure obligations of any benefit provided to the informant and disclosure of other information relevant to the informant’s credibility (725 Ill Compiled Stat. 5/151–21 (2019)). In 2017, Texas enacted a similar law. At the same time, Texas rejected proposals to ban jailhouse informants in death penalty cases. In 2009, Texas prohibited convictions based solely on the testimony of a jailhouse informant without corroborating evidence. A jailhouse informant who was bipolar and subsequently recanted contributed to Texas’s wrongful conviction and wrongful execution of Cameron Todd Willingham (Natapoff, Reference Natapoff2022: 95).

A number of states have enacted laws requiring prosecutors to maintain databases of jailhouse informants and to disclose their previous testimony and any benefits requested or received in relation to their testimony. In 2023, North Carolina enacted a law requiring the recording and disclosure of dealings between jailhouse informants and the state. California enacted a law requiring warnings to jurors about the potential unreliability of evidence from a jailhouse informant and requiring corroboration from a source other than another jailhouse informant (Reference Natapoff2022: 96–97). The range of regulation confirms how federalism can promote innovation and allow for comparative research within a nation-state.

In England, covert informers are regulated under the Regulation of Investigatory Powers Act enacted in 2000. Like PACE, it establishes various Codes of Practice and also establishes an Investigatory Powers Commissioner and the Investigatory Powers Tribunal to hear complaints. The use of informers must be approved by a senior police officer and placed on a registry. The number of such authorizations granted to the police has declined in recent years, raising concerns about the use of “off the books” informers. The English law has the virtue of recognizing that all informers, and not just jailhouse informers, may have incentives to lie in order to assist the police and prosecution. In Chapter 5, it will be suggested that the overriding characteristic of the English response to wrongful convictions has been the enactment of legislation.

The German Code of Criminal Procedure requires judicial approval of undercover investigations that are directed at particular persons, and either prosecutorial or judicial authorization is also required under the French Code (Bradley, Reference Bradley2007). The European approach generally subjects informants to a higher degree of regulation than in North America. Most of the remedied wrongful convictions in the European Registry involving perjury involve complainants and false confessions, not informers who lied, and this is a fertile area for additional research.

3.3.3 Executive Regulation

In Canada, most prosecutorial services have formed committees of senior prosecutors to vet the use of jailhouse informants. About a half of services keep a registry of such witnesses (Campbell, Reference Campbell2018: 118–130; FPT, 2018: 90). Manitoba prosecutors are subject to policies banning the use of jailhouse informants as the sole witness or if the informant has a previous perjury conviction. These policies have been voluntarily adopted by Canadian prosecutors in response to three public inquiries that have found that testimony from jailhouse informants has been a cause of wrongful convictions. Canadian prosecutors have considered but rejected a ban on the use of jailhouse informants (FPT, 2018: 90). These policies for vetting jailhouse informants before a prosecutor calls them as a witness have not been extended to accomplices and other unsavoury witnesses who may be inclined to lie, even though the courts have recognized the need for warnings to juries in such cases (R. v. Khela 2009 1 SCR 104).

3.3.4 Judicial Regulation

German courts have held that the use of a jailhouse informant to elicit information from a person subject to pre-trial detention was inadmissible because such attempts to obtain information were an abuse of the purpose of pre-trial detention (Weigend, Reference Weigend and Bradley2007: 259). On the other hand, Italy has allowed the use of incentivized witnesses through its pentiti system. An Italian journalist, Enzo Tortora, was wrongfully convicted on the basis of false statements by multiple jailhouse informants under the Italian system of granting immunity to witnesses to facilitate organized crime prosecutions. Tortora was held in jail for seven months in 1983 and convicted and sentenced to ten years of imprisonment. He received a pardon so that he could serve in the European Parliament in 1985, and he received subsequent acquittals by the Naples Court of Appeal and the Italian Supreme Court (Luparia, Cagossi, and Pittiruti, Reference Luparia, Cagossi, Pittriruti and Robins2023: 109).

In a case involving the prosecution of the notorious union leader Jimmy Hoffa, the United States Supreme Court allowed a jailhouse informant with incentives to lie to testify because the witness was subject to cross-examination and the judge gave a jury a warning about accepting his evidence (Hoffa v. US 385 U.S. 293 (1990); Thomas, Reference Thomas2008: 35–36). The Sixth Amendment right to counsel limits the ability of the state to direct jailhouse informants, but even in such cases, evidence otherwise inadmissible can be used for impeachment purposes. The Supreme Court has, however, rejected arguments that the testimony of jailhouse informants is always unreliable, instead deferring to the ability of the jury to make determinations of credibility (Kansas v. Ventris 556 U.S. 586 (2008)).

A Federal Court decision in 1998 applied a general offence against giving people benefits in exchange for their testimony to a prosecutor who offered a co-accused in a drug conspiracy case a deal that involved non-prosecution. An appeal court clarified the next year that the offence did not apply to prosecutors (US v. Singleton 165 F.3d 1297 (10th Cir. 1999), cert denied 527 U.S. 1024 (1999)). Dissenters argued that non-prosecution agreements “create a powerful incentive to lie and derail the truth-seeking purpose of the criminal justice system” (1999: 1310). Nevertheless, the American criminal justice system might have ground to a halt if prosecutors were not able to offer deals in exchange for giving evidence. The exemption of prosecutors from obstruction of justice offences on the basis of “sovereign immunity” fits into patterns of American extra-legalism where high state officials are often granted immunity from the rule of law (Roach, Reference Roach2011: ch 4; Trump v. United States, 603 U.S. 593, 2024).

At the same time, promises made to witnesses and even attempts by witnesses to obtain benefits for their testimony should be disclosed by prosecutors to the accused (Giglio v. United States, 405 U.S. 150, 1972; Wearry v. Cain 577 U.S. 385, 2016). The implicit premise is that disclosure should be requested and used to cross-examine incentivized witnesses and that cross-examination will reveal whether the witness is lying. At the same time, disclosure requirements only apply at trial and are not engaged when, as in the vast majority of cases, the accused pleads guilty (United States v. Ruiz 536 U.S. 622 (2002)). Again, even important rules such as disclosure that are designed to prevent wrongful convictions are subject to exceptions that threaten to swallow the rule.

The Supreme Court of Canada, in a 2000 decision, upheld a child murder conviction in a case where the state used testimony from a jailhouse informant. A majority indicated that it was an error for the trial judge not to give a specific warning that jailhouse informants may not be reliable, but one judge from that majority joined dissenters who argued that such warnings were not necessary in upholding the conviction (R. v. Brooks [2000] 1 SCR 237). One irony in the case is that the accused had not requested such a warning, presumably because it must be accompanied by a review of any evidence that confirmed the informant’s testimony. There is social science evidence suggesting that judicial warnings may often not have the desired effect and can even have counterproductive effects (Simon, Reference Simon2012: ch 6).

The next year, Peter Cory, a retired Supreme Court of Canada judge conducting the public inquiry where three jailhouse informants were used to shore up a weak murder case that resulted in a wrongful conviction, asked his former colleagues to reconsider the previous decision (Cory, Reference Cory2001). Alas, the Canadian courts have not, despite jailhouse informants featuring in 15 per cent of Canada’s remedied wrongful convictions. Courts have continued to accept evidence from jailhouse informants with an uptick of reported cases in recent years (Roach, Reference Roach2023a: 216, 329). The Australian High Court has taken a similar approach and ruled that warnings to juries about the danger of accepting the evidence of jailhouse informants are sufficient (Pollitt v. R. [1992] HCA 35).

3.3.5 Summary

The case for bans on the testimony of jailhouse informants is stronger than with respect to eyewitness identifications because it would be unusual and, in some jurisdictions, illegal for the prosecutor’s case to be based solely on the testimony of a jailhouse informant (Covey, Reference Covey2012; Roach, Reference Roach2007b). Yet all three branches of government have been unwilling to implement such categorical bans. In addition, the data from the registries suggest that notorious incentivized witnesses such as jailhouse informers in North America or “supergrasses” in England are only a small subset of witnesses whose mistakes or lies contribute to remedied wrongful convictions.

Some American state legislatures have been active in regulating the testimony of jailhouse informants, including with respect to disclosure of benefits received by jailhouse informants to the defence. Canadian prosecutors who are not elected have voluntarily adopted procedures to vet jailhouse informants (Campbell, Reference Campbell2018: 118–130), but this has not prevented their increased use in trials in recent years (Roach, Reference Roach2023a: 214–217). This raises concerns that some forms of reforms offered as an attempt to prevent wrongful convictions may actually be legitimating and enabling (McBarnet, Reference McBarnet1981). Jailhouse informants are regulated as the most flagrant and notorious example from the larger category of unreliable witnesses.

Accomplice and jailhouse informant testimony appears to play a significantly lesser role in remedied wrongful convictions recorded in both the British and European registries. There are some exceptions, including at least one Italian wrongful conviction related to the use of pentiti who received lesser sentences or no prosecutions for testifying against the accused in organized crime cases. There is a need for more research into why a snitching culture that inevitably produces unreliable evidence may be more prevalent in North America than in Europe. One possible explanation is increased regulation of informants in Europe (Ross, Reference Ross, Chesterman and Fischer2009). Another is differences in penal culture and the role of harsh mandatory sentences in North America that place intense pressure on the accused to provide state evidence in exchange for a sentence or charge bargain or a non-prosecution agreement.

3.4.1 Evidence from the Registries

False confessions were made in 16 per cent of the first 250 DNA exonerations in the United States (Garrett, Reference Garrett2011a: 18) and 460, or 12.6 per cent, cases out of 3,649 cases listed in the American registry as of the end of 2024. Garrett (Reference Garrett2011a: 38) reports that 33 per cent of DNA exonerees who falsely confessed were juveniles, including the Exonerated (Central Park) Five, and 43 per cent were mentally ill or disabled. Of the 460 false confession cases in the American registry, 180 involved teenagers.

False confessions were a factor in 98, or 20 per cent, of 494 cases listed in the UK registry as of the end of 2024. With some exceptions, the original convictions tended to come from the 1970s and 1980s, though some more recent cases have been overturned in part because police did not follow legislative guidelines under the PACE, 1984, for the taking of statements. More recently, remedied wrongful convictions based on false confessions have frequently involved expert evidence suggesting that the accused was suggestible.

False confessions were a feature of 14 or 16 per cent of 89 cases in the Canadian Registry of Wrongful Convictions, with half of the cases being classified as coerced confessions, two cases as voluntary confessions, and five cases where the accused denied making any confession.

The European Registry reveals that false confessions were a cause in 50 of 136, or 37 per cent, of remedied wrongful convictions. This number is quite high given that it has often been thought that inquisitorial systems place less weight on confessions than adversarial systems and that European police do not generally apply the guilt-presumptive Reid method of interrogation used in North America.

3.4.2 Legislative Regulation

The problem of false confessions came to attention in England in the 1972 Confait murder case. Three youths, two with borderline intelligence and a third who did not speak English as a first language, falsely confessed after police interrogations without their parents or another appropriate adult being present. After much investigative reporting, the Court of Appeal quashed the convictions on a reference by the Home Secretary (R v. Lattimore (1975) 62 Cr App R 53). The case and subsequent inquiry have been credited for provisions in the PACE Act enacted in 1984 that provided for the recording of interrogations and special protection for vulnerable persons (Poyser, Nurse and Milne, Reference Poyser and Milne2018: 22–23).

Code C of PACE stresses the importance of recording interviews and any statements. An appropriate adult must be present when vulnerable persons or young people are interviewed (Code C: s.11.15), subject to prescribed urgency exceptions (Code C: s.11.18). There are also requirements that medical personnel certify a detainee as fit to be interviewed, including evaluating risks to the detainee’s mental and physical health (Code C: 12.3). There are provisions to ensure detainees receive adequate rest and breaks in questioning every two hours (Code C: 12.8). Lawyers can, at the suspect’s request, attend police interviews with their clients subject to certain conditions (Code C: s.6.8). As with eyewitness identification, PACE, enacted in 1984, represented an early and comprehensive legislative framework that also contemplated additional executive and judicial regulation.

Many, but not all, American states have enacted laws that require the recording of statements. In 2008, North Carolina enacted a law requiring custodial interrogations to be recorded. The purpose of the law was defined as to “eliminate disputes about interrogations, thereby improving the prosecution of the guilty while affording protection to the innocent and increasing court efficiency” (Article 8 of Chapter 15A of General Statutes s.15A-211). Recording laws can protect the police from false claims of abuse. Legislative regulations of the immediate causes of wrongful convictions are more likely to be adopted if they offer benefits to police and prosecutors.

Most recording laws do not mandate automatic exclusion of unrecorded confessions. For example, in Illinois, unrecorded statements can be admitted if they are voluntary and reliable. At the same time, Florida, Georgia, South Carolina, Mississippi, Kentucky, Louisiana, Arkansas, Pennsylvania, Wyoming and Arizona have no legislation requiring any interrogations to be recorded. Federalism allows for innovation and experimentation but also laggard states.

A Missouri law provides that non-recording can be a basis for cuts in the funding to a police service that does not in good faith attempt to comply with the recording obligation (Missouri Statutes 590.70). The spending powers of governments can be used to encourage executive reforms that may reduce the risks of wrongful convictions.

An increasing number of American states have enacted laws since 2021 to prevent police use of deception when interrogating young suspects. As in the case with some state bans on teenage informants examined earlier, this takes a status-based approach, and it remains to be seen whether the regulations will ever be extended to adults. Under a 2022 Illinois law prohibiting the use of deception in interrogations of those under eighteen years of age, any use of deception means that a confession is presumed to be inadmissible unless the state can prove that it was voluntary (Juvenile Court Act s.5–401.6). Only a few states mandate exclusion when deception is used. These laws underline how reforms adopted in one American state can quickly migrate to other states, thus re-enforcing the idea that federalism can be a laboratory for experimentation. They are supported by the high percentage of false confessions made by teenagers and suggest that some, but not all American laws, are finally catching up with England’s 1984 PACE law.

The Argentinian Code of Criminal Procedure attempts to limit interrogations to those conducted by judges, but the police frequently record “spontaneous declarations” made by the suspect. Such extrajudicial statements are only accepted as circumstantial evidence if corroborated by other evidence. Statements to the police that the accused can prove are involuntary are inadmissible, but the police can follow leads from such statements. Similar rules and practices apply in Mexico. Suspects can be held up to forty-eight hours before making more formal statements to prosecutors, and such time is sometimes used to “soften-up” suspects. Defence counsel can be present when the prosecutor takes statements, and the police cannot ask questions during this stage. In practice, statements will only be excluded if the suspect proves they are involuntary (Bradley, Reference Bradley2007: 27–33, 364–370).

Germany allows both informal and formal investigatory questioning, but the latter triggers warnings that the suspect need not cooperate, that they can consult counsel, and they can demand the consideration of exonerating evidence. Counsel are, however, often not appointed until formal charges are made. Intentional misrepresentations by the police are prohibited, and courts have excluded evidence because the police lied in telling the suspect that they had overwhelming evidence of guilt or because the suspect had not been told of the right to counsel or right to silence (Weigend, Reference Weigend and Bradley2007: 256–261). Article 136a of the Code of Criminal Procedure forbids obtaining confessions “by ill-treatment, induced fatigue, physical interference, administration of drugs, torment, deception or hypnosis.” At the same time, there are a number of remedied wrongful convictions in Germany where the accused have falsely confessed to obtain a lighter sentence (European Registry, Andreas Boehmann). In one case, Holger Gensmer, a man with a mental disability, falsely confessed to the murder of a six-year-old child and was imprisoned for sixteen years despite his alibi and was then subsequently denied compensation on the basis that he contributed to his wrongful conviction (European Registry, Holger Gensmer). Another mentally disabled man, Ulvi Kulac, falsely confessed in 800 pages of interview transcripts to the murder and sexual abuse of a nine-year-old child and was imprisoned for eleven years (European Registry, Ulvi Kulac). Yet another case involved a German man with a mental disability who was imprisoned for thirty-one years after a false confession to the murder and sexual abuse of a seven-year-old child (European Registry, Dirk K.).

Italy does not allow even spontaneous statements to be used as evidence against a suspect before they are given right to counsel and right to silence warnings. The prosecutor can question suspects but must, in the absence of exigent circumstances, give their defence counsel twenty-four hours’ notice of the questioning. Despite this, there have been at least two cases of remedied wrongful convictions caused by false confessions in Italy, with one case involving a suspect who was mentally ill and another involving police pressure and deception (Luparia, Cagossi and Pittiruti, Reference Luparia, Cagossi, Pittriruti and Robins2023: 105). The Amanda Knox case involved an interrogation without counsel or an interpreter.

Taiwan’s Code of Criminal Procedure provides for recording of confessions and that “an accused shall be examined in an honest manner, violence, threat, inducement, fraud, exhausting examination or other improper means shall not be used.” Evidence obtained in violation of this rule can be excluded. Nevertheless, Taiwan has remedied wrongful convictions featuring an unrecorded false confession obtained after a six-hour interrogation. The judge in that case concluded the failure to record the interrogation was a “minor error” (Chin and Lin: 224–225).

India’s new criminal procedure law enacted at the end of 2023 will be examined in greater depth in Chapter 10.5, where the dominant role of false confessions in remedied Indian wrongful convictions will also be examined. One notable feature of the new law is that while the accused is supposed to be able to consult with a lawyer when “arrested and interrogated by the police,” they do not have the right to have the lawyer “present throughout the interrogation” (Bharatiya Nagarik Suraksha (Second) Sanhita, 2023 s.39). Section 183(1) makes it discretionary and not mandatory for magistrates to video and audiotape any confessions made by suspects in their presence. A new Indian evidence act maintains the inadmissibility of involuntary confessions caused by threats or inducements by persons in authority but also now specifically provides for the admissibility of confessions

This demonstrates that legislation can be used not only to restrict and regulate interrogation practices but also to make potentially unreliable statements admissible. As will be discussed in Chapter 10.5, the 2023 law may increase false confessions by increasing the time suspects may be detained in police custody.

As in India, false confessions obtained by police abuse are a leading cause of remedied wrongful convictions in China. China’s Criminal Procedure Law, as first amended in 2012 and again in 2018, mandates the exclusion of evidence as a means to protect against involuntary confessions. Article 53 of the 2018 law provides that “the use of torture to extract confessions is strictly prohibited as are threats, enticement, trickery and other illegal methods of gathering evidence; no person may be compelled to prove his own guilt.” Article 56 appears to apply an automatic exclusionary rule with respect to evidence obtained in violation of Article 53 when it states: “Criminal suspects or defendants confessions gathered by the use of torture to extract confessions or other illegal methods or witness testimony or victim statements gathered by the use of illegal methods such as violence or threats, shall be excluded.”

Writing in 2014, Renmin law professor He Jiahong (Reference He2014: 494) warned that while the exclusionary rule was a positive development, there was still a long way to go in ensuring that its promise on paper was implemented on the ground. He pointed to his own research on fifty remedied wrongful convictions, pointing out that while torture was possible in forty-seven of them, only four were acknowledged by courts and prosecutors as actually involving torture. He also noted that the use of torture to obtain confessions had been prohibited in China at least since 1979. Other research suggests that the exclusionary rule has not changed police practices in China (Lin, Chen and Wang, Reference Lin, Chen and Wang2019), with a 3 per cent exclusion rate at trial and on appeal (Kielsgard, Reference Kielsgard2020: 401). The Chinese experience suggests that the effects of legislative regulation designed to prevent false confessions will depend in large part on police acceptance and compliance with the legislation.

3.4.3 Judicial Regulation

Even before Miranda, the United States Supreme Court excluded involuntary confessions often obtained through torture or lynching threats directed toward Black suspects (Brown v. Mississippi 297 US 278, 1936; Chambers v. Florida 309 U.S. 227, 1940). In 1966, the Court in its famous Miranda decision required right to counsel warnings, no interrogation if a person subject to custodial detention requests counsel and attendance of counsel at the interrogation. The Court has, however, subsequently held that Miranda was not violated when the police told the suspect that they had no way of giving the detainee a lawyer, but one would be appointed later in court (Duckworth v. Egan 492 US 195, 1989). The accused does not have to be told of the matter under investigation (Colorado v. Spring 479 U.S. 564, 1987) or even that a lawyer is waiting to see him (Moran v. Burbine, 475 U.S. 412, 1986). Accused persons must clearly invoke their rights (Davis v. US 512 US 452, 1994). Brandon Garrett (Reference Garrett2011a: 37) reports that, like most suspects, all of the DNA exonerees who made false confessions waived their Miranda rights.

Although the American police cannot use threats or offers of benefits to obtain a confession, they can lie to the suspect (Frazier v. Capp 394 US 731 (1969); Oregon v. Mathiason 429 US 492 (1977)). Some state courts, however, have found that extreme forms of police deception may make a confession inadmissible (State v. Baker 465 P.3d 860, 2020). The United States Supreme Court has held that courts should not attempt to determine whether a confession is reliable when determining whether it has been voluntarily provided (Colorado v. Connelly 479 US 157, 1986). The case involved a mentally ill person who made a confession to a police officer. The US Supreme Court overturned lower court decisions that the confession was involuntary, ruling that reliability was not a constitutional issue. This creates a danger that false confessions will be obtained from suspects with mental disabilities or mental health issues. As of 2022, 69 per cent of the false confessions in the American registry involved suspects that had mental disabilities or mental illness, with only 8 per cent of false confessors “having no known mental disability or illness” (National Registry, 2022). American courts rejected a challenge to Brendan Dassey’s confessions despite evidence that the sixteen-year old had mental disabilities when he confessed to assisting his uncle Steven Avery in a rape/murder and despite evidence from videotapes that the detectives who interviewed him fed him some of the answers in his confession. Dassey v. Ditmann 877 F.3d. 297 (7th Cir, 2017) cert denied United States Supreme Court, June 25, 2018.

Texas courts encouraged the recording of interrogations as early as 1982, long before Texas enacted a statute to that effect in 2017 (Ragan v. State, 642 S.W.2d 489, 1982). Alaska was one of the first states to require recording of custodial statements as a result of decisions by the Alaska Supreme Court in 1980 and subsequently in 1985 (Mallott v. State, 608 P2d 737, 1980; Stephan v. State, 711 P2d 1156, 1985). The Court noted that recording will not only protect the accused but also the police from false claims of abuse. The Alaska courts were somewhat more assertive than legislatures in stressing that the exclusion of unrecorded confessions would be necessary to ensure that the police would comply with the recording obligations. Other courts in New Jersey, Minnesota and Massachusetts established similar regimes (State v. Scales, 518 N.W.2d 587, 592, 1994; Commonwealth v. Diaz, 661 N.E.2d 1326, 1996).

Canadian courts have also strongly suggested that police should record interrogations, but neither they nor the legislature requires recording. The Supreme Court of Canada in a 2000 case recognized the dangers of false confessions that could cause wrongful convictions (R. v. Oickle 2000 SCC 38). The Court took note of published social science even though Canadian courts had resisted allowing experts to testify about false confessions (Trotter, Reference Trotter2003). At the same time, the Court concluded that most false confessions were caused by improper police techniques involving threats or inducements. This needs to be reconsidered given data indicating that most false confessors have mental disabilities or mental health issues. Subsequent to the Oickle decision, voluntary false confessions to serial sexual assaults made by a severely mentally disabled person, Simon Marshall, were revealed, but only after he had been imprisoned and abused in prison for five years after he pled guilty to sexual assaults that he did not commit (Roach, Reference Roach2023a: 11–15).

The Supreme Court subsequently indicated that only extreme forms of trickery or deception could render a confession involuntary and that counsel does not have a right to attend interrogations (R. v. Sinclair, 2010 SCC 35). Canadian courts have accepted confessions made after prolonged and persistent interrogations and despite the suspects’ frequent statements that they do not want to talk to the police (R. v. Spencer, 2007 SCC 11). They have rejected a challenge to the use of the guilt-presumptive Reid method of interrogation on the basis that courts should concern themselves with voluntariness and oppression and not what interviewing techniques the police use (R. v. Ordonio, 2025 ONCA 135). They have also maintained a minimal operating mind requirement with respect to the necessary mental capacity of the suspect, neglecting research on how mental health issues and addictions can result in false confessions (Sherrin, Reference Sherrin2005). The only requirements to protect the vulnerable from false confessions are statutory requirements that those under eighteen years of age have a “reasonable opportunity” to have counsel, parent or another appropriate adult attend the interrogation (Youth Criminal Justice Act. S.C. 2002 c.1, s.146). This is a status-based approach that does not apply to adults with cognitive issues that may lead to false confessions.

Canada is infamous for its use of Mr. Big stings, where undercover police officers target suspects for elaborate stings. Such stings are migrating to other jurisdictions, including Australia and New Zealand. The stings often offer the suspect the financial benefits of being part of a criminal organization in exchange for an eventual confession about the crime being investigated to a police officer posing as the leader of the organization (the so-called Mr. Big). This technique was devised as a way to short-cut both constitutional rights to counsel and confession rule regulation because neither applied because the target of the sting did not know that he was being questioned by police officers. This raises a recurring issue that the police can frequently avoid regulation in one area by using less regulated investigative techniques. At the same time, the often lengthy and unrecorded stings provided the police with plenty of opportunities both to provide threats and inducements to the target and to provide hold-back information that could give any false confession made by the suspect a degree of credibility. The Mr. Big technique was used an unknown number of times but was associated with at least two remedied wrongful convictions (Canadian Registry, Kyle Unger and Andrew Rose). It has also been involved in the wrongful murder conviction of David Lyttle in New Zealand (White, Reference White and Robins2023b: 211).

In 2014, the Supreme Court of Canada announced that henceforth statements taken by the police during a Mr. Big sting would be presumed to be inadmissible unless the state established their reliability. The Court defended this approach as striking “a just balance” between responding to the “risk of wrongful convictions that stem from false confessions” while also ensuring that “the police are not deprived of their opportunity to use their skill and ingenuity in solving serious crimes” (R. v. Hart, 2014 SCC 52: Para 3). The Court excluded the confession in this case, but most Canadian courts have admitted evidence that is still obtained through such schemes, generally relying on the presence of corroborating evidence (Iftene and Kinnear, Reference Iftene and Kinnear2020). It is difficult to know whether those subject to such stings are being made aware of hold-back information and whether they have vulnerabilities that could lead to false confessions. The expensive stings are generally only used in stalled murder investigations, which also underlines the dangers of confirmation bias and noble cause corruption (MacFarlane, Reference MacFarlane2006).

False confessions are more likely to be obtained in countries where the police believe they are free or justified in using torture or other threats. In one case from the Bahamas, the police obtained a confession from a twenty-year-old suspect to robbery after six days of police detention. The accused alleged he was hit by the police with a baseball bat and had a plastic bag placed over his head. He maintained his innocence but was convicted at trial without any legal representation and on the basis of his confession. He was acquitted of another robbery that he had confessed to after a lawyer who represented him in that case raised a reasonable doubt about the voluntariness of his confession in police custody. He was only acquitted on appeal from conviction by the Judicial Committee of the Privy Council after twelve years of imprisonment on the basis of what the Board said was a confession that should never have been admitted (Ariste v. The King (The Bahamas) 2023 UKPC 18 [37]). The Privy Council has stressed the importance that the police observe restrictions on interrogations imposed by the Judges Rules (Peart v. the Queen (Jamaica) 2006 UKPC 5).

3.4.4 Executive Regulation

Many Canadian police services have voluntarily decided to record interviews, and some have decided no longer to train officers to use the guilt-presumptive Reid technique (FPT, 2018: 66–69). Brandon Garrett (Reference Garrett2010: 1115) has argued that “police procedures are a crucial piece of any reform effort” and noted that police disclosure of hold-back information to suspects is discouraged under the Reid method of interrogation. He has also proposed police use different interrogation techniques with respect to those who are vulnerable to suggestion, including youth and those with mental disabilities. This proposal is well supported by the data from both DNA exonerations and registries of wrongful convictions, including both the American and European registries.

Since 1992, the British police have used Planning and Preparation, Engage and Explain, Account, Clarify and Closure, and Evaluation (PEACE) interviewing techniques that they developed with assistance from the Home Office and researchers. PEACE stresses more open-ended information gathering rather than accusatorial techniques often used in North America. At the same time, PEACE allows for persistent questioning even when suspects invoke the right to silence. The Criminal Justice and Public Order Act, 1994, heightened the leverage of the police by allowing an accused’s silence to be used against him or her as evidence.

PEACE has also been imported to other countries including New Zealand, Norway, Germany and some Australian states. Prosecutors have raised concerns about importing PEACE to Canada because, unlike in the UK, the accused’s refusal to respond to police questions is not admissible as evidence against the accused (FPT, 2018: 63). That said, training detectives to use more open-ended and less harsh interview techniques and also to take special care when interviewing vulnerable suspects has a potential to reduce false confessions. The Broward County Sheriff’s Office in Florida, after a series of false confessions from mentally disabled suspects in high-profile cases, adopted a new policy that requires annual training on identifying and questioning such suspects. The policy requires notification of a superior officer before interrogations and, where possible, the provision of an appropriate adult to be present during the questioning. If a confession is obtained, it must be vetted by a unit supervisor or if there is no corroborating evidence by another investigator, prosecutor and a psychologist (Leo, Reference Leo2008: 312–313).

Like the other interventions examined earlier, executive regulation can be combined with other regulation. The PACE approach combines standards for interviewing devised by both the legislature and the executive and enforced by the judiciary. It is thus what Keith Findley (Reference Findley2016) refers to as a hybrid approach that is likely the most effective. Only seven of the ninety-eight remedied wrongful convictions in the United Kingdom involving false confessions occurred in the 2000s, and three of these cases were guilty plea cases resulting from the Post Office scandal (British registry). One of these cases involved a UK Supreme Court decision from Scotland that applied a European Court of Human Rights decision stressing the need for a lawyer during police interrogations (Cadder v. Her Majesty’s Advocate [2010] UKSC 43).

Policing laws or regulations could require the police to address any of the causes of wrongful convictions and provide for some ongoing inspection to determine the adequacy of the responses. Steps to make the police response public as well as the regular publication of statistics and other updates on the police response could also facilitate continued scrutiny by civil society and the media.

3.4.5 Summary

Recording of interrogations is the most popular form of regulation to attempt to prevent false confessions and has been adopted in different jurisdictions by all branches of government. It is generally not tied to an automatic rule that unrecorded confessions will be inadmissible. More research is needed to determine the degree to which recording helps reveal and exclude false confessions at trial. Recording has been accepted in no small part because it also offers advantages to the police by protecting them against false claims of abuse. There is a danger that a recorded confession, even if false, will be more readily accepted at trial (Ibusuki, Reference Ibusuki2019).

Interviewing style may be an important factor in false confessions and depends in large part on executive self-regulation by the police. Some American states have recently enacted legislation designed to prevent the use of police deception in interrogations of those under eighteen years of age. This may be a start but is underinclusive, and PACE makes special allowance for all suspects known to be vulnerable. This is supported by evidence from the American registry that the vast majority of those who falsely confessed had mental disabilities or mental illnesses (National Registry, 2022). The English police have, since 1992, used interrogation techniques that are not as guilt presumptive and do not rely on deception or minimization as much as the Reid technique, which is still frequently used in North America. These changes build on both legislative regulation and judicial enforcement. They may help explain why, since the 2000s, false confessions have not been a factor in most remedied wrongful convictions in England.

3.5.1 Evidence from the Registries

Garrett (Reference Garrett2011a: 89) reports that forensic evidence was presented in 74 per cent of the first 250 DNA exonerations. This high rate of forensic error was related to the fact that hair comparison and/or blood typing were frequently used in the minority of cases where biological evidence was present. DNA exoneration cases may not be typical and may over-represent the importance of forensic error in all wrongful convictions. Nevertheless, they provide an important reminder of the fallibility of the forensic sciences.

As of the end of 2024, the American National Registry reports 1,048, or 29 per cent of 3,649, remedied wrongful convictions involved false or misleading forensic evidence, this lower percentage reflecting that the registry is not limited to the minority of cases with DNA evidence. Of these cases, 144 involve drug analysis, but there are also over 32,000 people listed in the group’s registry who had convictions vacated and charges dismissed because of two scandals in drug labs in Massachusetts.

The British registry has 81 of 494, or 16.4 per cent, of remedied wrongful convictions as of the end of 2024 that involved false or misleading forensics, with only 7 of those cases involving affirmative evidence of innocence. Eighteen of the seventy-nine cases involved women, generally convicted of killing babies and very young children in their care, a topic that will be examined in greater depth in Chapter 8.6.

The Canadian registry has a much higher percentage than either the American or British registries of thirty-one or 35 per cent of eighty-nine wrongful convictions involving false or misleading forensic evidence. This reflects both the relatively small size of the Canadian registry and also ten exonerations relating to the testimony of one discredited forensic pathologist, Charles Smith (Roach, Reference Roach2023a: chs 2 and 4). Three of the seven Canadian public inquiries that have examined wrongful convictions have examined problems with various types of forensic science evidence (Roach, Reference Roach2009). In addition, two other inquiries focused on flawed drug testing analysis that produced two criminal wrongful convictions of women as well as false and misleading expert testimony in child protection proceedings (Roach, Reference Roach2023a: 88–93).

The European registry has 34, or 25 per cent of 136, of its recognized exonerations relating to false or misleading forensic evidence as of the end of 2024. Several of these cases involved flawed forensic pathology evidence about the cause of death, with subsequent evidence indicating that the cause of death may have been accidental or natural (European Registry, Lucia de Berk). Other cases (European Registry, Tomasz Komenda; European Registry, Viggo Kristianson) involved hair and bite marks in a rape murder that were later contradicted by DNA evidence.

3.5.2 Judicial Regulation

Judicial decisions to admit expert opinion evidence are implicated in most wrongful convictions that include flawed forensics as one of their immediate causes. In a comparative study of admissibility standards in England, the United States, Australia and Canada, Edmond et al. (Reference Edmond, Cole, Cunliffe and Roberts2013) have concluded that judges in all those countries have admitted forensic science of unknown or dubious reliability, including fingerprints, bitemark analysis and identification of video images and voice recordings. Although most American jurisdictions apply Daubert 509 U.S. 579 (1993), which is supposed to require that expert evidence has been tested with error rates being determined, this judicial precedent has “made little discernable difference to practice” (Edmond et al., Reference Edmond, Cole, Cunliffe and Roberts2013: 39). Other commentators have suggested that Daubert gives American judges the power to demand more reliability from the forensic sciences but “they have only rarely availed themselves of this power” (Koehler, Mnookin and Saks, Reference Koehler, Mnookin and Saks2023). Judges not only routinely allow forensic experts to provide opinion evidence of unknown reliability, but they also assume that adversarial cross-examination will reveal its shortcomings (Edmond et al., Reference Edmond, Cole, Cunliffe and Roberts2013: 99). Courts have been reluctant to exclude incriminating forensic evidence presented by the state.

The United States Supreme Court has required that the accused be able to confront and cross-examine the person who actually conducted the forensic analysis that was the basis for the expert opinion (Melendez-Dias v. Massachusetts, 557 U.S. 305, 2009; Bullcoming v. New Mexico, 564 U.S. 647, 2011). The right to confrontation is a potential safeguard, but also one that is diluted by the lack of access to qualified defence experts and restrictive tests for determining ineffective assistance of defence counsel. Cross-examination will also not help in the vast majority of cases that result in a guilty plea (Garrett, Reference Garrett2021: 135–136).

In July 2024, the United States Supreme Court refused to hear a case where a 1985 murder conviction was based on bitemark evidence. Justice Sotomayor dissented and argued that bitemark evidence has “now been roundly condemned by the scientific community and retracted by the expert who introduced it at his trial” (McCrory v. Alabama, 144 S.Ct. 2483, 2024). She also noted that six states including Texas have introduced writs based on changed science (Beety, Reference Beety2022), underlining that American legislatures have been more open to allowing evidence of changed science to disrupt the finality of convictions than American courts.

There are special challenges with respect to forensic expert evidence in developing countries. In a case from Bahamas, the Privy Council rejected the accused’s argument that an expert must attend at trial to be cross-examined. The expert refused to attend because of passport and security concerns, and the prosecution made no arrangements for testimony by video link (Stubbs v. The Queen (Bahamas) [2020] UKPC 27 [109]). In a case from Bermuda, a police officer was qualified to give expert opinion evidence about gangs (Myers v. The Queen [2015] UKPC 40). Police are also allowed to testify as experts in other countries, including Canada, but often with judicial restrictions on the claims that they can make (R. v. Sekhon 2014 SCC 15). There are also many Caribbean death penalty cases where fresh expert evidence obtained from outside the islands is offered on appeal (Brown v. State (Trinidad and Tobago) 2012 UKPC 2).

A Canadian inquiry into a series of wrongful convictions caused by faulty forensic pathology highlighted that in many of these cases guilty pleas were entered after defence counsel could not find experts to challenge the now discredited testimony offered by the pathologist who was the state’s witness and in the state’s employ. It recommended increased legal aid funding for defence experts after finding a case in which a middle-class family mortgaged their home to pay for defence experts who successfully rebutted the state’s pathologist evidence. Effective oversight of the pathologist’s testimony could have prevented many wrongful convictions. The inquiry recommended enhanced oversight of the work of forensic pathologists (Goudge, Reference Goudge2008; Roach, Reference Roach2009).

Courts in both England and Canada have resisted attempts to challenge the foundational validity of forensic sciences such as fingerprint analysis that is based on the experience of the practitioner. The English Court of Appeal in one case even stated that it hoped “that the courts will not be troubled in the future by attempts to rely on published work by people who have no practical experience in the field and therefore cannot contradict or bring any useful evidence to bear on issues that are not always contained in scientific journals” (R. v. Weller [2010] EWCA 1085 at para 38). They also have accepted police as experts on video identifications without adequately exploring the grounds of their expertise or their impartiality (Edmond and Wortley, Reference Edmond and Wortley2024). Canadian courts have held that a trial judge who took judicial notice of published research about the lack of foundational validity of fingerprinting had erred and also upheld evidence of a single fingerprint match used to convict a person after expert evidence challenging the validity of fingerprinting was heard at trial (Cunliffe and Edmond, Reference Cunliffe and Edmond2021).

In 2011, the Law Commission of England proposed a new legislative standard for the admissibility of expert evidence that would stress the need for reliability. No legislation was enacted, but Criminal Procedure Rules were changed to encourage experts not to stray from their area of expertise and to provide sufficient information to allow courts to determine if their expert evidence was sufficiently reliable to be admissible as evidence. Practice directions also encouraged expert witnesses to detail their engagement with peer-reviewed research, to situate their expert opinion in the range of relevant expert opinion and to confirm their impartiality. The move to greater use of peer-reviewed research has generally been praised, but it also introduces a new potential source of error. One Canadian inquiry found that misinterpretation of published research helped explain a Canadian wrongful conviction on the basis of hair and fibre comparison that was subsequently corrected by DNA (Roach, Reference Roach2009). Some of the current controversy over Lucy Letby’s murder and attempted murder convictions involved claims by the author of published research that this research has been misinterpreted (The Guardian, 2025a).

In 2015, the Supreme Court of Canada altered the test for the admission of expert opinion evidence after examining three public inquiries involving wrongful convictions and forensic error. In addition to long-standing requirements that such expert opinion evidence must be relevant, necessary, given by a properly qualified expert and not be subject to any exclusionary rules, courts were instructed that even when these requirements were satisfied to balance the probative value and prejudicial effects of the expert evidence (White Burgess 2015 SCC 23). As with the leading American admissibility decisions, this decision was made in civil litigation where parties often have more resources to challenge the admissibility of expert evidence. In any event, the new Canadian test has encouraged some courts to edit expert opinions when convinced that certain opinions may have greater prejudice than reliability (R. v. France 2017 ONSC 2040). At the same time, such judgments require defence lawyers and judges who understand the science sufficiently to make such determinations. The Canadian Supreme Court has stressed that experts must be impartial but indicated that an employment relation alone (e.g., with the police or the government) or even a reasonable apprehension of bias will not generally render an expert partial and unqualified (White Burgess 2015 SCC 23). The Supreme Court of Canada has found that a police officer had strayed beyond their area of expertise when allowed to testify that he had never seen a case where a drug courier did not know the drugs they were carrying (R. v. Sekhon 2014 SCC 15).

In Argentina, the court appoints expert witnesses from a list of official experts, but the parties can also appoint their own experts at their own expense (Bradley, Reference Bradley2007: 48). In Mexico, the prosecutor or the court can appoint expert witnesses from a list, and they are automatically qualified to give opinion evidence. The defence can also propose its own expert, but they must be accepted by the judge as qualified (Reference Bradley2007: 388). In Germany, the prosecutor often appoints an expert during the investigation who is then usually appointed by the court as a trial expert. The defence must not only pay for but also demonstrate that another expert is required because of deficiencies in the court’s expert (Reference Bradley2007: 267). The German law professor Thomas Weigend has noted that “theoretically, the German system of court-appointed expert witnesses should make for a neutral, detached role of experts, in contrast to the ‘hired gun’ syndrome in adversarial systems.” At the same time, Professor Weigend (Reference Weigend and Bradley2007: 267) notes, “defence counsel complain that experts tend” to favour the prosecution.

Edmond and Vuille (Reference Edmond and Vuille2014) have also found similar problems of unreliable forensic science in Switzerland as in England, despite Switzerland’s use of investigating magistrates who appoint the expert. The Swiss system “invests considerable trust in past practice, the competence and impartiality of the commissioned and the abilities of the deciding magistrate,” who is free to assign whatever weight he or she wishes to the forensic evidence under inquisitorial systems of free proof (Reference Edmond and Vuille2014: 236). Second opinions can be requested under Swiss law, but may require the party asking for them to pay for them, thus casting them as “hired guns” (Reference Edmond and Vuille2014: 237). This study concluded that the Swiss approach to the reliability of forensic science was not appreciably better than that in England (Reference Edmond and Vuille2014: 257). Even though defence experts are now allowed in many inquisitorial systems, they can be marginalized as compared to the more official expert appointed by the court or the prosecutor. This also suggests that the frailties of forensic evidence are a universal immediate cause of wrongful conviction, though the regulation of forensic service providers may differ from jurisdiction to jurisdiction.

3.5.3 Legislative Regulation

This area confirms a pattern seen with respect to other immediate causes of wrongful convictions, with American state legislatures experimenting with different reforms. In 2015, California enacted a law that allows post-conviction challenges to convictions on the basis of scientific advances or repudiations by the expert who testified at the criminal trial (Cal Penal Code s.1473). In 2018, Connecticut and Wyoming enacted similar laws that also allow evidence of new guidelines governing forensic evidence to be entered as new evidence (Conn Pub. Act 18–61; W.S. 1977 s.7-12-401). Nevada enacted a similar law in 2019 to allow new science to be introduced when relevant to issues of factual innocence (NRS c.34, s.176.515). West Virginia added new science to habeas corpus relief in 2021 (SWV s.53-4A-1).

One American DNA expert, Eric Lander, complained in 1989 that “clinical laboratories must meet higher standards to be allowed to diagnose strep throat than forensic labs must meet to put a defendant on death row” (Murphy, Reference Murphy2015: 283). Fortunately, some progress has been made. Texas created a Forensic Science Commission in 2005 in the wake of its wrongful conviction and execution of Cameron Todd Willingham on the basis of indications of arson used at his trial that have subsequently been shown to occur in accidental fires. The majority of the commissioners are university-based scientists joined by two forensic scientists, a prosecutor and a defence lawyer. Governor Rick Perry, who had denied Willingham clemency, fired three of its members, but the Texas Commission was still able to release its critical report about the fire “science” in the Willingham case but without discussing whether there was misconduct in that particular case (Thompson and Casarez, Reference Thompson and Casarez2016: 719). This illustrates some tension between post facto accountability and efforts to improve standards in the future. The Texas Commission’s mandate has been expanded by legislation to include accreditation, licensing and discipline of those providing forensic services. Brandon Garrett (Reference Garrett2021: 203–204) reports that thirteen states have created similar bodies and that Texas has ordered audits of previous uses of hair, bite mark and DNA evidence.

Despite a decision holding that access to DNA evidence is not constitutionally required (District Attorney v. Osborne 129 S.Ct. 2308, 2009), all American state legislatures and Congress have enacted laws providing some post-conviction access to DNA evidence. These laws are not always matched with laws or policies that require the retention of evidence that may be subject to DNA testing or laws that provide quality control with respect to DNA testing. There has also been more emphasis on expanding DNA databases than on analysing crime scenes and retention of evidence (Murphy, Reference Murphy2015). In addition, some DNA laws restrict access to DNA testing to those who are still in custody or convicted of certain offences or impose time bars. In Ohio, DNA testing is not available if the accused pled guilty (Medwed, Reference Medwed2022: ch 7). Errors in the interpretation of DNA, especially mixed samples, trace evidence and contamination, have also caused wrongful convictions (Gill, Reference Gill2014).

In 2018, the Canadian province of Ontario enacted the Forensic Laboratories Act in response to concerns about labs that provided inaccurate forensic results as a response to wrongful convictions and child protection proceedings based on faulty drug testing by the Motherrisk lab, which was associated with a prestigious hospital but not accredited for forensic purposes (Roach, Reference Roach2023a: 223–224). The Act requires forensic labs to be accredited by the International Organization for Standardization (ISO) with respect to matters such as codes of conduct and proficiency testing. At the same time, these quality controls have been criticized for lacking “real clarity or teeth” (Garrett, Reference Garrett2021: 154). In response to wrongful convictions caused by forensic pathology, Ontario also amended its Coroner’s Act to require the Chief Forensic Pathologist to keep a register of qualified forensic pathologists and be subject to an oversight council.

Forensic error has played an important role in many high-profile English wrongful convictions, including the Birmingham Six and Guildford Four case and the Sally Clark and Angela Cannings baby deaths case. In 2012, the government-owned and operated Forensic Science Service was disbanded with forensic services being privatized or brought into police services. The English Forensic Science Regulator relied upon voluntary codes of conduct. It was not until 2021 that it was given a statutory basis when a private member’s bill first introduced in 2018 became law in the form of the Forensic Science Regulator Act, 2021. The English Forensic Science Regulator now has powers to investigate and issue compliance notices with respect to forensic work performed after 2023, again indicating some tensions between responding to past failures and improving standards in the future. Failure to abide by its codes of conduct may be admissible in evidence. This depends on adequate and informed defence lawyering.

Amoako and McCartney (Reference Amoako and McCarthy2022) have described an English shift from a voluntary “carrot” based approach to a “stick” based compliance approach. They argue that it raises capacity concerns about the ability of small forensic services to comply. This could lead to more police domination of the provision of forensic sciences. Interaction between all forensic service providers and the police raises concerns about confirmation and investigative bias. The Reference He2021 law fits into a pattern to be discussed in Chapter 5 of England relying on national legislation in its responses to miscarriages of justice.

In 2023, India enacted a new criminal procedure law that, as will be discussed in greater depth in Chapter 10.5, requires forensic investigation in serious cases. This law responds to India’s low conviction rate as well as well-publicized impunity concerning sexual crimes. That said, it raises capacity issues and introduces new sources of error. Studies of forensic labs in India have raised concerns that they are understaffed and do not devote enough resources to validation or quality control. There have been calls for India to create a Forensic Science Regulator on the English model (Project 39A, 2023).

3.5.4 Executive Regulation

Australian law professor Gary Edmond (Reference Edmond2016) has contrasted the way that scientific organizations test the reliability of forensic science with the way that courts test their reliability. The courts often rely on the experience and qualifications of the experts, prior decisions, adversarial challenges, judicial warnings to jurors and judicial editing of the expert opinion. In contrast, scientific organizations rely on empirical tests to determine the validity and accuracy of the methods used, error rates, proficiency testing of individual experts, compliance with standards and best practices, peer review, multidisciplinary acceptance, transparency and attempts to control for contextual bias and confirmation bias. Professor Edmond notes that there is almost no overlap between legal and scientific approaches to testing reliability and suggests that the latter are superior. Such conclusions support Peter Neufeld’s (Reference Neufeld2005) blunt argument that forensic science reform is unlikely to come from the courts and needs to take place “upstream” in the organizations that produce forensic science and have the skills necessary to perform the type of validity and proficiency testing that Edmond identifies with scientific approaches to determining reliability.

Brandon Garrett (Reference Garrett2021: ch 13) similarly has argued for crime labs to reform themselves by increased proficiency testing and quality control, use of error rates, blind testing, increased disclosure and transparency, firewalls from investigators and greater scientific supervision of police collection and retention of evidence. He describes how a Houston crime lab that contributed to many wrongful convictions reformed itself through such reforms. Legal process analysis can include executive reforms, and there is some evidence that this is essential with respect to forensic science. One challenge will be the fragmentation of many distinct disciplines in forensic science and the dangers of relying on a slow and reactive system of reforming forensic science discipline-by-discipline and jurisdiction-by-jurisdiction (Roach, Reference Roach2009).

The Obama administration created a National Commission on Forensic Science in 2013, which made recommendations on a variety of issues, including proficiency testing and the communication of results, but it was allowed to expire by the Trump Administration. In 2015, the President’s Council of Advisors on Science and Technology (PCAST) issued a report that found scientific validity in DNA analysis but with some reservations about complex mixture samples. It found bitemark, firearms and footwear analysis to be subjective and without scientific validation. Fingerprinting was also subjective but had more validity but also required steps to ensure proficiency and guard against confirmation bias.

Alas, the PCAST report received pushback from the FBI and Obama’s Attorney General Loretta Lynch, who stated that the Department of Justice would not accept its recommendations because they remained “confident that, when properly used, forensic science evidence helps juries identify the guilty and clear the innocent” and that “current legal standards regarding the admissibility of forensic evidence are based on sound science and sound legal reasoning” (Shniderman, 2016). The Trump administration similarly urged courts not to give credence to the concerns raised in the PCAST report and its call for increased use of error rates. This demonstrates deep bipartisan pushback to reforms that ask for the courts to be more restrictive in terms of the admissibility of forensic science. This is unfortunate because reforms generated by forensic science providers, while necessary, may have to be re-enforced by the courts (Findley, Reference Findley2016).

There have been a number of wrongful convictions in the Caribbean related to faulty forensics. In 2013, Lescene Edwards was convicted of murdering his ex-wife despite a suicide note being found at the scene. In 2022, the Judicial Committee of the Privy Council overturned the murder conviction on the basis of new forensic evidence both related to the cause of death and expert evidence at trial that failed to detect gunshot residue on either the accused or the deceased. The Committee accepted that Edwards did not have funds at trial to obtain such expert evidence. His limited funds at trial were spent on trying to refute the prosecution’s arguments that he had forged the deceased’s suicide note. It also noted that the costs of defence experts were high because there were no independent experts in Jamaica that could provide the evidence (Edwards v. The Queen (Jamaica) 2022 UKPC 11 [45]).

A murder conviction from Bermuda was overturned by the Judicial Committee of the Privy Council because of concerns about errors made in trace DNA evidence presented at trial. The Director of Public Prosecutions agreed that the errors made by the Florida-based DNA lab were so widespread that there would be an audit of all 247 cases where the lab’s results had been used in cases in Bermuda. The Board also observed that the American expert who bolstered her testimony with misleading statistics did not attest that she understood the duties of expert witnesses to be impartial, which differed from her duties as an expert witness in the United States, where she usually testified. The Board declined a request that new experts eventually obtained by the defence participate in the audit of past DNA cases on the basis that it cannot require the DPP to conduct a review in a particular way or by a particular person (Washington v. The King (Bermuda) 2024 UKPC 34 [58] [63]).

3.5.5 Summary

In many ways, regulatory responses to forensic error lag behind those to the other three immediate causes of wrongful convictions discussed in this chapter. This relates to the fact that, as the National Academy Science (2009) report pointed out, many forensic sciences used daily in criminal courts lack scientific validity and often have inadequate quality controls. Even if work is conducted properly in the lab, miscommunication of the results in courts can result in wrongful convictions, with some studies finding greater error rates in the oral testimony of experts than in their original reports. There is a need for a hybrid approach that combines legislative regulation and oversight of forensic science, internal reforms adopted by forensic science providers, and decisions by courts that encourage forensic service providers to invest in needed research, testing and quality control and that ensure that experts do not stray beyond their area of expertise or make exaggerated claims.

Judges decide whether to admit expert opinion evidence, but the evidence of the effectiveness of judicial gatekeeping in preventing judicial error is not robust. The United States Supreme Court has stressed the importance of testing, error rates (Daubert 509 U.S. 579, 1993) and cross-examination of experts (Melendez-Dias v. Massachusetts, 557 U.S. 305, 2009), but much depends on the capabilities of defence counsel. Moreover, trial safeguards are not even theoretically applicable in the vast majority of cases that end in guilty pleas.