In this paper, I study ability peer effects among teenagers. The identification relies on a fuzzy regression discontinuity approach where assignment into high-ability classes constitutes the source of identifying information. An important feature of this system is that both types of classes are taught by the same teachers, they follow a common curriculum and take the same exams. Students are in general unaware of the system prior to school start as it is unofficial. In cases where they are informed of the system's existence, they do not know where the threshold lies and school switching possibilities are limited. I find significant and sizable effect on the academic achievement of students around the assignment threshold. Being assigned to a high-ability class increases academic achievement, measured by year grade and spring exam results, by 0.47 and 0.32 standard deviations, respectively.