It is widely held that similarity of attitudes promotes interpersonal attraction. However, Rosenbaum (1986a) argued that the relationship between attitudinal similarity and attraction is actually a relationship between attitudinal dissimilarity and repulsion. Contrasting predictions of the similarity-attraction and dissimilarity-repulsion hypotheses were tested in a main within-subjects experiment. Ninety subjects first judged a stranger, a randomly sampled same-sex university student, in a no-attitude information control condition. Later they judged the same student again, knowing that the stranger shared 0.00, 0.50 or 1.00 proportion of similar attitudes with them (N = 30). As predicted by the similarity-attraction hypothesis, both similar and dissimilar attitudes affected attraction. Moreover, the effects of similar and dissimilar actitudes were contingent upon the level of similarity of attitudes assumed by the subjects in the no-attitude information control condition. In an auxiliary between-subjects experiment, attraction response was also higher in the experimental condition of similar attitudes (N = 19) than in the control condition of no-attitude information (N = 20). These results reaffirm the similarity-attraction relationship but reject the dissimilarity-repulsion hypothesis. In addition, they call attention to the proper consideration of assumed similarity in Byrne's (1971) reinforcement-affect model.