We estimate the extent of ability peer effects and explore the mechanisms through which they operate. Using within-school variation in the proportion of low-ability students in Israeli schools, we find that the proportion of low-ability peers has a negative effect on the performance of regular students. An exploration of the underlying mechanisms show that, relative to regular students, lowability students report higher levels of satisfaction with their teachers. However, a higher proportion of low-ability students has detrimental effects on teachersÕ pedagogical practices and on the quality of inter-student and student-teacher relationships, and increases the level of violence and classroom disruptions.In this article, we study the existence, magnitude and mechanisms of peer effects on educational outcomes in middle and high schools. Specifically, we investigate whether having unusually low-achieving classmates has any effect on the educational outcomes of regular pupils and we also identify some of the mechanisms and channels through which these effects operate.Using studentsÕ exact date of birth and the rules governing school enrolment in first grade, we identify as potentially low-ability students those who were born substantially before the relevant threshold date for their cohort (ÔrepeatersÕ). 1 We show that repeaters have substantially lower academic achievements relative to both regular students and relative to any other group of students identified based on their socioeconomic background. We then assess how classroom ability composition, as measured by the proportion of repeaters in the grade, affects scholastic achievements of high school students. Lastly, we use a unique national survey administered to middle school students to identify whether peer composition affects teachersÕ pedagogical methods in the classroom, the level of disruption and violence, and the quality of inter-student and student-teacher interactions.The article makes two main contributions to the literature on peer effects. First, it identifies the effects of having low-ability peers on studentsÕ achievement using a methodology that is less susceptible to bias from potential confounders associated with peer quality and to a reflection problem. Second, it goes beyond the peer effect