We estimate the effect of neighbours' characteristics and prior achievements on teenage students' educational and behavioural outcomes using census data on several cohorts of secondary school students in England. Our research design is based on changes in neighbourhood composition caused explicitly by residential migration amongst students in our dataset. The longitudinal nature and detail of the data allows us to control for student unobserved characteristics, neighbourhood fixed effects and time trends, school-by-cohort fixed effects, as well as students' observable attributes and prior attainments. The institutional setting also allows us to distinguish between neighbours who attend the same or different schools, and thus examine interactions between school and neighbourhood peers. Overall, our results provide evidence that peers in the neighbourhood have no effect on test scores, but have a small effect on behavioural outcomes, such as attitudes towards schooling and anti-social behaviour.