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 over-estimates the returns to degrees with the highest average earnings and...