The German practice of compelling weak students to repeat a class has come under heavy criticism recently. Many observers fear that this practice is, at best, useless or even counterproductive. However, little is known so far on the consequences of having to repeat a class, as compared to be confronted with new course material in the next class.This paper, therefore, aims at generating empirical evidence on the effect of class repetition on individual educational attainment. Since an experimental study is precluded,we utilize an instrumental variable approach to control for unobserved heterogeneity between respondents.Our estimation results suggest that there exists a negative association between repeating a class and educational attainment. However, taking unobserved heterogeneity into account yields a statistically significant and quantitatively substantial positive effect of class repetition on educational outcomes.