Stereotype threat can be conceptualized as a situation involving the potential loss of status for a group member. Given that baseline testosterone levels have been shown to be related to status-relevant concerns and behavior in both humans and other animals, it was hypothesized that individual differences in baseline testosterone might moderate stereotype threat effects. In Experiment 1, highly math-identified male and female undergraduates completed a math test after answering questions that focused on gender stereotypes in math (stereotype threat for women) or coming to college (control). Before taking the test, however, all participants provided saliva samples allowing analysis of baseline testosterone levels. Testosterone levels interacted with participant sex and stereotype threat to affect performance on the math test. For women exposed to negative stereotypes of their group, increased testosterone levels produced poorer math performance, but there were no effects of testosterone level in the control condition. Men outperformed women overall, and scores were not affected by stereotype exposure or testosterone level. In Experiment 2, men highly identified with math provided saliva samples then completed a math test that was described as useful for identifying "only individuals who were 'exceptional' in math ability" (status-enhancement potential) or "only individuals 'weak' in math ability" (no status-enhancement potential). Males high in testosterone outperformed men low in testosterone in the status-enhancement condition. In contrast, there was no difference in performance between high and low-testosterone males when the framing offered no possibility of status-enhancement. These data lend support to the notion that stereotype threat effects occur due to status-relevant concerns, and that the specific content of stereotypes predict whether those concerns will augment or attenuate performance.