Citizenship is the most important right a host country can bestow on its immigrant population. Yet, little is known which citizenship policies work and who actually benefits from them. To answer these questions, we estimate the marginal returns to citizenship on children’s school performance and skill development. For identification, we use two national reforms, which facilitated naturalization for first-generation immigrants and introduced birthright citizenship. We find substantial unobserved heterogeneity in returns with reverse selection on gains, i.e., the returns are highest for those with the lowest propensity of take-up. Citizenship significantly improves the school performance of immigrant children but has only modest effects on test scores. Policy simulations indicate that raising citizenship take-up would generate sizable benefits overall. Based on marginal treatment response functions, we also show that expanding birthright citizenship carries higher returns than facilitating naturalization.