It is commonly perceived that increasing incentives improves performance. However, the reaction to increased incentives might differ between men and women, leading to gender differences in performance. In a natural experiment, we study the gender difference in performance resulting from changes in stakes. We use detailed information on the performance of high-school students and exploit the variation in the stakes of tests, which range from 5% to 27% of the final grade. We find that female students outperform male students in all tests—but to a relatively larger degree when the stakes are low. The gender gap disappears in tests taken at the end of high school, which count for 50% of the university entry grade.