Analyzing university faculty and graduate students data for ten of the top U.S. economics departments between 1987 and 2007, we find persistent differences in the gender compositions of both faculty and graduate students across departments. There is a positive correlation between the share of female faculty and the share of women in the PhD class graduating 6 years later. Using instrumental variable analysis, we find robust evidence that this relation is causal. These results contribute to our understanding of the persistent under-representation of women in economics, as well as for the persistent segregation of women in the labor force.