Social desirability is generally thought to underlie the propensity for survey respondents to tailor their answers to what they think would satisfy or please the interviewer. While this may in fact be the underlying motivation, especially on attitudinal and opinion questions, social desirability does not seem to be an adequate explanation for interviewer effects on factual questions. Borrowing from the social psychology literature on stereotype threat, we test an alternative account of the race-of-interviewer effects. Stereotype threat maintains that the pressure to disconfirm and to avoid being judged by negative and potentially degrading stereotypes interferes with the processing of information. We argue that the survey context contains many parallels to a testing environment in which stereotype threat might alter responses to factual questions. Through a series of framing experiments in a public opinion survey and the reliance on the sensitivity to the race of the interviewer, our results are consistent with expectations based on a theory of “stereotype threat.” African American respondents to a battery of questions about political knowledge get fewer answers right when interviewed by a white interviewer than when interviewed by an African American interviewer. The observed differences in performance on the political knowledge questions cannot be accounted for by differences in the educational background or gender of the respondents.