We analyze a matched employer-employee panel data set and find that female leadership has a positive effect on female wages at the top of the distribution, and a negative one at the bottom. Moreover, performance in firms with female leadership increases with the share of female workers. This evidence is consistent with a model where female executives are better equipped at interpreting signals of productivity from female workers. This suggests substantial costs of underrepresentation of women at the top: for example, if women became CEOs of firms with at least 20% female employment, sales per worker would increase 6.7%.