Three experiments investigated how awareness of stereotypes (i.e., metacognitive beliefs) affect performance in an intellectually threatening context. Study 1 involved undergraduate students (N = 77, 37 minority, 40 White) who indicated they were aware of the stereotype, Whites are perceived to be more intelligent than stigmatized minorities in society. Participants were told they would complete several computerized tasks, including two verbal analogy tests which were described as highly predictive of performance on other intelligence tests (stereotype threat for minorities.) Participants completed an initial verbal analogy test, meant to elicit anxiety about the second test, and reported their level of anxiety. They completed a working memory task (i.e., a prime) that raised either confidence or doubt in their abilities and completed a questionnaire to check for stereotype threat. They did not actually complete the second verbal analogy test. Minority participants experienced more stereotype threat and anxiety. Anxiety predicted decreased working memory when the task primed doubt, but not when it primed confidence, for both minority and White participants. In Study 2, female participants (N = 111) completed a math test and a working memory task. In the stereotype threat condition, participants were told that the test was highly predictive of ability and that their scores would be compared with mens (stereotype threat for women). In a control condition, participants were told that the test was a problem-solving exercise and that the study would compare individual performances. Participants then reported their level of anxiety, completed the same working memory task as in Study 1, priming either confidence or doubt, and completed a questionnaire to check for stereotype threat. Women in the stereotype threat condition reported more concern about how their gender would be evaluated, but they did not experience higher anxiety. Women who reported more anxiety and were under stereotype threat had lower working memory scores when they were primed with doubt, but not when primed with confidence. Study 3 measured anxiety by comparing measures of salivary alpha-amylase (sAA) activity. Female undergraduates preparing to take the GRE (N = 43) took a survey measuring self-relevance of performance on each of the GRE sections and gave a saliva sample to assess baseline sAA. 1-3 days later, participants took a reappraisal tendency survey (measuring the tendency to regulate negative emotions) and then took a practice GRE test, providing saliva samples before each of the two test sections. The participants completed a questionnaire asking them to rate their self-doubt. Initial arousal (high sAA) predicted decreased performance only on the math section (stereotype-relevant task) and only among people with low reappraisal tendency. For those with high math sAA, high reappraisal tendency predicted lower math doubt. These studies show how awareness of stereotypes can created anxiety and doubt that affects the ability to perform well under stereotype threat.