Recent studies for primary and secondary education find positive effects of the share of girls in the classroom on achievement of boys and girls. This study examines whether these results can be extrapolated to post-secondary education. We conduct an experiment in which the shares of girls in workgroups for first year students in economics and business are manipulated and students are randomly assigned to these groups. Boys tend to postpone their dropout decision when surrounded by more girls, and there is also a modest reduction in early absenteeism. On the other hand, boys perform worse on courses with high math content when assigned to a group with many girls. Overall, however, we fail to find substantial gender peer effects on achievement. This in spite of the fact that students' perceptions of the behavior of themselves and their peers are influenced by the share of girls.