The present study examined stereotype challenge and threat responses through a manipulation of social comparison information. Female participants received feedback about an ostensibly matched partner who was either similar (i.e., female) or dissimilar (i.e., male) to themselves. After learning that they would be taking a math test that either has or has not demonstrated gender differences in the past, participants learned that their partner either performed very well or very poorly on the test. Subsequently, participants took the math test themselves, and their performance was evaluated. It was predicted that interacting with a peer whose performance was consistent with the negative stereotype would exacerbate stereotype threat effects, while interacting with a peer whose performance challenged the negative stereotype would eliminate stereotype threat effects. The results support these predictions, highlighting the manner whereby peer-group social comparisons can exert a powerful influence on women’s math test performance when negative stereotypes are salient.