We study the effect of exposure to immigrants on the educational outcomes of U.S.-born students, using a unique dataset combining population-level birth and school records from Florida. This research question is complicated by substantial school selection of U.S.-born students, especially among White and comparatively affluent students, in response to the presence of immigrant students in the school. We propose a new identification strategy, comparing sibling outcomes with the inclusion of family fixed effects, to partial out the unobserved non-random selection into schools. We find that the presence of immigrant students has a positive effect on the academic achievement of U.S.-born students, especially for students from disadvantaged backgrounds. Moreover, the presence of immigrants does not negatively affect the performance of affluent U.S.-born students, who typically show a higher academic achievement compared to immigrant students. We provide suggestive evidence on potential channels.