This paper evaluates how sharing a kindergarten classroom with low-achieving repeaters affects the long-term educational performance of regular first-time kindergarten students. Exploiting random assignment of teachers and students to classes in Project STAR, I document three sets of causal impacts: students who are exposed to repeaters (1) score lower on standardized tests at the end of kindergarten, an effect that fades out in later grades; (2) show persistent improvements in non-cognitive skills such as effort and discipline; and (3) are more likely to graduate from high school and to take a college entrance exam around the age of eighteen. I show that the positive spillovers from repeaters on long-term educational attainment are likely driven by the differential accumulation of non-cognitive skills by repeater-exposed students during childhood. The improvements in these skills are in turn a result of behavioral adjustments by teachers, students, or parents to the presence of low-achieving repeaters in the classroom.