Three studies investigated the authenticity of prejudice-related discrepancies. A comprehensive discrepancy questionnaire was developed (Study 1), which yielded small as well as large discrepancy scores. Study 2 indicated that discrepancy scores were stable, and personality could not account for the relation between discrepancies and their affective consequences. In Study 3, low-prejudice participants responded to jokes about Blacks under high or low distraction. Behavioral validation for self-reported discrepancies was found, such that participants with larger discrepancies evaluated the jokes more favorably under high than low distraction, but participants with smaller discrepancies provided equally unfavorable evaluations in both distraction conditions. Implications for understanding people's abilities to avoid potentially prejudiced responses and their self-insight into such abilities are discussed.