For over twenty years, a voucher system has been used in Chile to promote competition in the educational system between public and private schools. Attending a private subsidized school is associated with increased standardized test scores, but the apparent impact is relatively small. Controlling for school choice using a supply-side instrument (school availability at community level) implies substantially larger impacts of the voucher system. The effect of parents' education on academic performance is smaller than that implied by simple OLS estimates that do not control for school choice. Finally, the results also show that family school choice is gender biased, females are sent more often to voucher schools while males are sent more often to private (non voucher) schools. In addition, the TSLS estimates show that females gain less than males from going to voucher schools.