Two experiments were designed to distinguish different kinds of stereotype threat based on their source. Specifically, the study addressed whether stereotype threat that either focused on the group (group-threat) or on the performance of the individual group member (self-threat) might occur in different situations and be influenced by different factors. To test these possibilities, female undergraduates in Experiment 1 completed a difficult math test in one of four conditions: when the test was described as non-diagnostic of their ability (non-diagnostic control), when it was described as diagnostic of their abilities (traditional stereotype threat), when it was diagnostic but that they would grade their own test and results would be anonymous (self-threat), or when the test was diagnostic and would be used to assess average performance of men and women with little regard to their individual score (group-threat). Results showed that (after controlling for Math SAT scores) performance was worse in the traditional stereotype threat and the self-threat conditions compared with the control condition. Performance in the group-threat condition was also poorer than in the control condition, but not to a statistically significant degree. It was speculated that the weaker results in the group-threat condition might have occurred because only women who strongly identify with their gender might show performance decrements when their group identity was threatened. Experiment 2 assessed this possibility by having women complete the test in a non-diagnostic control, self-threat, or group-threat condition, but, prior to doing so, had them complete a measure of gender identification. Test performance was worse in the self-threat and group-threat conditions than in the control condition. Moreover, the strength of gender identification affected women's performance in the group-threat but not in the self-threat or control condition; women who most strongly identified with their gender performed most poorly in the group-threat condition. These findings suggest that stereotype threat can arise either because individuals are concerned about the applicability of a gender stereotype to themselves and to their group. Susceptibility to stereotype threat regarding one's group, however, appears to be moderated by the strength of one's identification with that group.