This paper asks how a student's classmates affect her returns to college. We exploit a “tracking” admission system at a selective Colombian university that led to large differences in mean classmate ability for students in the same programs. In a regression discontinuity design, we find that...