This paper uses a unique database from a French university to analyze gender biases in student evaluations of teachers (SETs). The results of generalized ordered logit regressions and fixed-effects models suggest that male teachers tend to receive higher SET scores because of students' gender biases. Male students in particular express a strong bias in their favor: male students are approximately 30% more likely to give an excellent overall satisfaction score to male teachers compared to female teachers. The different teaching dimensions that students value in men and women tend to correspond to gender stereotypes. The teaching dimensions for which students perceive a comparative advantage for women (such as course preparation and organization) tend to be more time-consuming for the teacher, compared to the teaching dimensions that students value more in men (such as class leadership skills). Men are perceived as being more knowledgeable (male gender stereotype) and obtain higher SET scores than women, but students appear to learn as much from women as from men, suggesting that female teachers are as knowledgeable as men. Finally, I find that if women increased students' continuous assessment grades by 7.5% compared to the grades given by their male colleagues, they could obtain similar overall satisfaction scores as men. Yet, women do not act on this incentive (men and women give similar continuous assessment grades), suggesting that female teachers are unaware of students' gender biases. These biases have strong negative consequences for female academics, who may spend more time on teaching to try to obtain high SET scores, reducing time available for research. The results suggest that better teaching is not necessarily measured by SETs.