Girls in India have significantly lower school enrollment rates than boys. Anecdotal evidence suggests that intra-household gender discrimination is the most important reason for this gap, but empirical support in most of the previous literature is rare and seemingly inconsistent with patterns in related economic research. I propose that this may be due to a combination of endogeneity issues and inadequate attention to age-specific forms of discrimination. I analyze school enrollment using a plausibly exogenous source of income variation for rural households: rainfall shocks. The results show that girls’ school enrollment is more vulnerable to rainfall shocks than that of boys, with 6-10 year old children driving these effects. This empirical pattern is consistent with young girls being out of school because of credit constraints, and older girls being out of school because of low perceived net benefits of education.