Public support for Brexit has declined since the 2016 referendum. We argue that part of this decline is due to cohort replacement where many older voters (who support Brexit) have passed away, while younger voters (who oppose Brexit) have entered the electorate. Using a series of original YouGov surveys from 2016, 2018, 2020 and 2022, each representative of the UK electorate, we first demonstrate the large and stable differences in Brexit support between younger and older voters. Next, we employ demographic decomposition calculations to estimate that cohort replacement alone accounts for approximately one third of the decline in aggregate Brexit support in just 6 years (with two thirds of the decline being explained by within‐cohort changes). Furthermore, by combining our data on Brexit support with Office for National Statistics cohort projections up to 2030, we derive testable hypotheses about the pressure that cohort replacement will continue to put on Brexit support over the next decade across a wide range of potential scenarios. Altogether, our study demonstrates the powerful role that cohort replacement plays in shaping British (and European) politics in the post‐Brexit world.

Cohort replacement – the replacement in a population of older cohorts by their successors who developed under different conditions – is an important process behind cultural change. Research on public opinion indicates that a large proportion of aggregate change is the result of cohort replacement rather than of individuals changing their minds. However, some publicly salient issues, like gay rights, appear to be exceptions. Why different issues show different patterns of change is not well understood. In this paper, we investigate whether opinions on sensitive – that is, hard to discuss – issues might change differently than opinions on less sensitive issues. We use data from the 1981–2020 World Values Surveys and newly collected data on the sensitivity of survey items to compare aggregate changes in public opinion on 56 survey items in eight countries. Our key finding is that survey items on more sensitive issues seem to change more through cohort replacement.