In all three studies described in this article, novices were trained to associate traits from the DSM-IV Cluster B personality disorder categories with four letters of the alphabet. Novices were also taught various gender associations (i.e., "base rate" information) with the categories (none, stereotype-consistent, or stereotype-inconsistent). Results showed that when no gender associations were taught (Study I), case gender did not affect ratings of case vignettes. When associations were consistent with social stereotypes (Study II), case gender influenced ratings in the expected direction. When associations were the opposite of social stereotypes (Study III), only the ratings for narcissistic and histrionic were affected. Across the three studies, these results suggested that subjects rated cases in accordance with known "base rates," but that the correspondence between base rates and stereotypic associations affected the consistency and magnitude of this base rate effect.