Without doubt the validity of scientific theories and their usability for solving societal, economic, ecological, and health-related problems are contingent on the existence of robust and replicable empirical findings. However, a review of the recently published replication literature portrays a rather pessimistic picture of the replicability of even very prominent empirical results, as is evident both from large-scale meta-analytic replication projects and from distinct attempts to replicate selected examples of well-established key findings of personality and social psychology. The present chapter offers a twofold explanation for this undesirable state of affairs. On one hand, the widespread evidence on replication failures reflects to a considerable extent the neglect of logical and methodological standards in replication studies, which sorely ignore such essential issues as manipulation checks, reliability control, regressive shrinkage, and intricacies of multi-causality. On the other hand, however, the community of behavioral scientists must blame themselves for an intrinsic weakness of their corporate identity and their incentive and publication system, namely the tendency to mistake sexy and unexpected findings for original insights and neglect of the assets of cumulative science and theoretical constraints.