Social identity theory and related viewpoints hold that an individual's social group memberships become part of the psychological self, with diverse and powerful effects on thoughts, feelings, and behavior. However, the successful tests of this hypothesis to date have mainly involved evaluative judgments or reward allocations as dependent measures. Adapting a method developed by Aron et al. (1991), based on the same logic as Stroop interference, we provide more direct evidence that an in-group can be represented as part of the self. Subjects made speeded self-descriptiveness judgments for a variety of traits. Analyses showed that responses were slower and involved more errors for traits on which the individual believed he or she differed from an in-group, compared to traits on which the individual and the in-group were perceived as similar. Matches or mismatches between the self and a salient out-group had no effect. This evidence suggests that cognitive representations of the self and an in-group are directly linked, to the point where reports about the self are facilitated for traits on which the self and in-group are similar, and inhibited for traits involving dissimilarity.