Investigated the effects of comparisons with similar and dissimilar performances on estimates of relative ability. Male and female undergraduates (n = 162) took a test of analogy-making ability, and received veridical feedback about their own performance and manipulated feedback about the scores of other college students. Supporting Festinger's similarity hypothesis, variations within a range of small discrepancies between self and others' scores produced significant changes in ability estimates, while variations within a range of large discrepancies had no significant impact. This pattern of results only held for discrepancies relative to the modal score of others, and not relative to their highest or lowest scores. There were also indications that subjects with average performances were less influenced by the comparison feedback than were subjects with low or high scores. The discussion focused on the similarities and contrasts between seeking comparison information versus being influenced by it.