We study how gender differences in performance under competition are affected by the provision of information regarding rival's gender and/or differences in relative ability. In a laboratory experiment, we use two tasks that differ regarding perceptions about which gender outperforms the other. We observe women's underperformance only under two conditions: 1) tasks are perceived as favoring men and 2) rivals' gender is explicitly mentioned. This result can be explained by stereotype-threat being reinforced when explicitly mentioning gender in tasks in which women already consider they are inferior. Omitting information about gender is a safe alternative to avoid women's underperformance in competition.