We study the optimal design of student nancial aid as a function of parental income. We derive optimal nancial aid formulas in a general model. For a simple model version, we derive mild conditions on primitives under which poorer students receive more aid even without distributional concerns. We quantitatively extend this result to an empirical model of selection into college for the United States that comprises multidimensional heterogeneity, endogenous parental transfers, dropout, labor supply in college, and uncertain returns. Optimal nancial aid is strongly declining in parental income even without distributional concerns. Equity and eciency go hand in hand.