Racial affirmative action policies are widespread in college admissions. Yet, evidence on their effects before college is limited. Using four data sets, we study a U.S. Supreme Court ruling that reinstated affirmative action in three states. Using nationwide SAT data for difference-in-differences and synthetic control analyses, we separately identify the aggregate effects of affirmative action for whites and for underrepresented minorities. Using state-wide Texas administrative data, we measure the effect of affirmative action on racial gaps across the pre-treatment test score distribution. When affirmative action is re-instated, racial gaps in SAT scores, grades, attendance, and college applications fall. Average SAT scores for both whites and minorities increase, suggesting that reductions in racial gaps are driven by improvements in minorities' outcomes. Increases in pre-college human capital and college applications are concentrated in the top half of the test score distribution. [Support from the Lab for Economic Applications and Policy and the Connaught Fund.]