This study examined performance and effort expended in testing contexts under stereotype threat. Undergraduate students (N = 148; 69 women, 79 men) who had passed or received advanced placement or transfer credits for Calculus I were told they would complete a test under different instructions. Some participants were told that the test was about general problem-solving abilities (control), some were told it was a math test in which men typically outperform women (stereotype threat), and some were given this same instruction but also informed about the nature and anxiety-producing effects of stereotype threat (stereotype threat with attribution instructions). Performance accuracy on the test (number of correct answers/ number of problems attempted) produced no significant effects. However, women attempted fewer problems in the stereotype threat and stereotype threat with attribution instructions conditions compared with the control condition. In addition, women also attempted fewer problems than men in the stereotype with instructions condition. The authors speculated that the failure to find typical stereotype threat effects on performance accuracy might have occurred because stereotype threat was inadvertently triggered in all conditions by the collection of demographic data prior to testing. Even if so, this would not account, however, for why attribution instructions did not attenuate threat or why men would not have performed better than women across all conditions. Despite these ambiguities, the results do provide some evidence that stereotype threat can undermine task effort.