This paper analyzes the effects of selective schools on students’ educational and labor market outcomes. We utilize regression discontinuity design based on the centralized admission system of upper secondary schools in Finland to obtain quasi-random variation for selective high school offers and attendance. By using nationwide administrative data, we first show that the selective schools do not improve high school exit exam scores, even though there is a large jump in peer quality for students attending selective schools. Despite lacking short-term effects, we find that selective schools increase university enrollment and graduation in the long run. Yet, we do not observe positive effects on income. Importantly, our results suggest that selective high schools or better peer groups do not improve students’ human capital or skills, but affect their preferences on educational choices after the secondary school.