We hypothesized that, in a masculine task, only token women leaders who were empowered through position (by being appointed leader) and expertise (trained with task-relevant information) and legitimated by a male experimenter as credible would be more effective in influencing the performance of their all-male groups than appointed-only and appointed-trained leaders. Thirty women undergraduates each led a small group of male students on a moon survival task. The hypothesis was supported. Videotapes of group interactions revealed that appointed-trained leaders interrupted group members and used tentative tag questions in failed attempts to share their task-relevant knowledge. In addition, group members reported the most dissatisfaction with appointed-trained leaders who, without legitimacy, violated diffused gender roles by presuming to be expert on a masculine task. The importance of the organizational empowerment of token women is underscored.