This paper investigated whether different kinds of self-affirmation are differentially effective in protecting individuals from stereotype threat. Specifically, three studies assessed if affirmations highlighting on one's intrinsic self-worth (focusing on valued intrinsic self-aspects, such as unconditional relationships and important personal values) might be more effective than affirmations highlighting one's extrinsic self-worth (focusing on valued extrinsic self-aspects, such as conditionally accepting relationships and imposed social standards). In Experiment 1, male and female undergraduates were randomly assigned to complete sentence fragments referring to the self to provide an intrinsic (e.g., "being a ____ makes me feel ____") an extrinsic (e.g., "when I am a successful ____ I receive ____") or no affirmation (control; e.g., "watching television is a good way to ____"). After completing the affirmation manipulation, students completed a task involving mental arithmetic. Students who affirmed the self engaged in less self-handicapping before beginning the math task than in the other two conditions, and those individuals performed best on the math test (although comparisons with the control condition were not statistically significant). Experiment 2 focused on female students and utilized a manipulation of stereotype threat. Women completed either the intrinsic or extrinsic self-affirmation task from Experiment 1, then they were told they would complete a "Quantitative Examination" that either was "an indication of your mathematical intelligence" (stereotype threat), or involved "testing problems for future research" (control). Women in the stereotype threat condition also indicated their gender before attempting the exam. Analysis of exam performance revealed that women who intrinsically self-affirmed performed better on the exam in the stereotype threat compared with the control condition. However, women who extrinsically self-affirmed performed worse on the exam in the stereotype threat compared with the control condition. In Experiment 3, women who completed a task designed to provide an intrinsic or extrinsic affirmation received either positive or no feedback from a supposed evaluator. All women then completed a task that has been shown to reflect the accessibility of thoughts about rejection by others. Women who received no feedback had more thoughts about rejection if they had extrinsically self-affirmed than if they had intrinsically self-affirmed. Women who received positive feedback showed lowered accessibility of thoughts about rejection. These studies show that extrinsic self-affirmation increases concerns about rejection, augmenting the effects of threat that can arise when stereotypes are invoked. In contrast, intrinsic self-affirmation appears to lessen concerns about rejection, producing improved performance under stereotype threat.