We examine the impact of ethnic school composition on students’ educational outcomes using Swedish population register data. We add to the literature on the consequences of ethnic school segregation for native and immigrant students by distinguishing social interaction effects from selection and environmental effects through one- and two-way fixed effects models. Our findings demonstrate that native and immigrant students’ grades are relatively unaffected by social interaction effects stemming from the proportion of immigrant schoolmates. However, we find nontrivial effects on their eligibility for upper secondary school. Immigrants’ educational outcomes are weakly positively affected by the proportion of co-ethnics in school.