Accuracy and bias in self-perceptions of performance were studied in a managerial group-discussion task. Ss ranked their own performance and were ranked by the 5 other group members and by 11 assessment staff members. Although the self-perceptions showed convergent validity with the staff criterion, Ss were less accurate when judging themselves than when judging their peers. On average, Ss evaluated their performance slightly more positively than their performance was evaluated by either the peers or the staff; however, this general self-enhancement effect was dwarfed by substantial individual differences, which ranged from self-enhancement to self-diminishment bias and were strongly related to four measures of narcissism. Discussion focuses on issues in assessing the accuracy of self-perceptions and the implications of the findings for individual differences in self-perception bias and the role of narcissism.