This study analyzes the gender gap in self- and peer evaluations based on a laboratory experiment. Five players performed a creativity task in a high-stakes winner-takes-all tournament. The treatment without validation informed all players that evaluations that they will conduct determine who will win. The treatment with public validation additionally informed them that they can see an objective performance measure of all players (including themselves) at the end of the experiment which is irrelevant for winning. The results show that men give themselves better selfevaluations compared to women when there is no validation. This gender difference vanishes completely when providing public validation.