We study the effects of peers on school achievement, with detailed data on children making the same primary to secondary school transition in consecutive years in England. Our estimates show that secondary school composition, on entry at age 12, affects achievement at age 14, although the effect sizes are small. These secondary school peer effects originate in peer characteristics encapsulated in family background and early achievements (age 7), rather than subsequent test score gains in primary school. Our specifications control for individual unobservables and school fixed effects and trends, rendering peer group composition conditionally uncorrelated with student's characteristics.