A vast literature on the spillovers from girls’ education focuses on the impact of maternal education on child outcomes. This paper is the first to investigate whether externalities from investing in girls’ education may be realized earlier – before they have children of their own. In many developing countries, oldest sisters share significant child care responsibilities in the household and potentially play an important role in younger siblings’ learning. I propose a model incorporating the effect of the oldest sister that predicts competing effects of increasing oldest sister’s schooling on younger siblings’ human capital. Using an identification strategy that exploits the gender segregation of schools in Pakistan, I estimate the impact of the oldest sister’s schooling on the human capital acquisition of her younger brothers. I find that oldest sister's schooling has significant, beneficial impacts for younger brothers’ schooling, enrollment, literacy and numeracy. An additional year of schooling for the oldest sister increases the younger brother’s completed years of schooling by 0.42 years and his probability of being enrolled by 9.6 percent. It also increases the probability of a primary school-aged younger brother being literate and numerate by 7-19 percent. I discuss the implications of these results for policies targeting girls’ education. These findings indicate that evaluations of such policies that consider only effects on the girls and their children may underestimate their total benefits.