These experiments assessed the impact of gender-stereotype activation on performance expectations in negotiation situations, a domain in which gender stereotypes favor men. In Experiment 1, MBA students formed male-female dyadic pairs to negotiate over the purchase of a biotechnology plant. Half of the negotiation teams were told that the negotiation task was extremely diagnostic of negotiating ability (stereotype threat for women), and half were told the task was not an accurate gauge of negotiating ability (control). Men were expected to do better (and, in fact, did so) than did women in the threat condition, but there were no differences in expectations or performance in the control condition. In Experiment 2, one partner in female-male and male-male negotiation dyads were either told that effective negotiators are rational, assertive, and self-interested (i.e., traditional gender stereotypes of males designed to induce stereotype threat for women), or were given no information about traits (control). Results showed that men were more confident and more effective in negotiating with a woman, particularly when gender stereotypes had been invoked. Women's performance was no worse in the stereotype threat compared with the control condition, however, suggesting that women might have been operating under threat in both conditions. Experiment 3 employed a much stronger instantiation of gender stereotypes to test the hypothesis that stereotypes might at least in some circumstances lead to reactance, producing behavior in opposition to the activated stereotype. Male-female dyads completed the same negotiation used in Experiment 1, except that before the task, participants were told that traditional gender stereotypes were linked to differential success (weak stereotype condition), or that men tend to outperform women (strong stereotype condition). Women made stronger offers and were more successful than men in the negotiation only in the strong stereotype condition, suggesting that women sought to disprove the gender stereotype and did so successfully when stereotype activation was clear and strong. In Experiment 4, male-male and male-female dyads completed a negotiations with no mentioning of traits (control), with the strong stereotype instructions used in Experiment 3 (strong stereotype condition), or with instructions emphasizing the importance of a shared social identity (superordinate identification condition). Women in the strong stereotype condition achieved superior outcomes and showed evidence of distancing themselves from traditional female stereotypes. In contrast, men in that condition showed evidence of self-doubt and concern about failing to meet gender-based expectations of strong performance. These findings show that subtle manipulations of stereotype threat can undermine performance but that more explicit manipulations, at least in some individuals, can produce consequences in opposition to the activated stereotype.