This research examines whether reading a text presenting scientific evidence concerning the phenomenon of stereotype threat improves or disrupts women's performance in a subsequent math task. In two experimental conditions participants (N=118 ) read a text summarizing an experiment in which stereotypes, and not biological differences, were shown to be the cause of women's underperformance in math (Gender-relevant condition), or the deficits of Afro-Americans on verbal tests (Control condition). Results showed that, whereas men's performance was not affected by the information provided, women who properly understood the mechanism of stereotype threat had their math performance disrupted in the Gender-relevant condition. These findings suggest that the mere presentation of research evidence on stereotype threat, in the absence of other interventions aimed at reducing the aversive effects of negative stereotypes, may have harmful effects on the targets of stigma.