eople’s automatic preference between groups often determines their automatic preference between unknown individual members of these groups, a prominent example for automatic prejudice. What happens when the person making the judgment has long known the target individuals? Practice might automatize the deliberate judgment of the individuals. Then, if deliberate judgment is non-prejudiced, automatic prejudice might decrease. In 29 studies (total N = 4,907), we compared automatic and deliberate preferences between a famous member of a dominant social group and a famous member of a stigmatized social group. In most studies, we chose pairs based on prior self-reported preference for the member of the stigmatized group. Across all studies, automatic preference was discrepant from deliberate preference, often favoring the member of the dominant group. We replicated these results with various target individuals, two pairs of social groups (Black/White, Old/Young), two automatic evaluation measures, and in two countries (Studies 1-23). The automatic pro-dominant preference was stronger when visual characteristics of the group were present rather than absent (Studies 24-25), suggesting a stronger effect of group characteristics on automatic than on deliberate preference between the individuals. The automatic preference between the individuals was related more strongly to automatic than to deliberate group preference (Studies 26-27), yet it was still moderated by individuating information (Studies 28-29). Our results suggest that familiarity with members of a stigmatized group does not automatize the positive deliberate evaluation of these individuals, and does not dethrone group evaluation from its central role in the automatic evaluation of the individual.