This paper uses a college-by-graduate degree fixed effects estimator to evaluate the returns to 19 different graduate degrees for men and women. We find substantial variation across degrees, and evidence that OLS overestimates the returns to degrees with high average earnings and underestimates...