In 2012, the Brazilian government passed a law that mandated all federal higher education institutions to implement a 50 percent admission quota for historically disadvantaged students. I study the implications of this regulation on the academic performance of non-targeted students. Identification rests on the use of pre-law crosswise variation in specially admitted student representation to instrument for exogenous changes in the student body composition afterward. Increased enrollment of targeted students due to the affirmative action caused an increase in the variance of academic ability within university programs. However, I find no evidence that quota-students affect the dropout of non-quota students.