The self-esteem of 80 male subjects was temporarily either raised or lowered by giving them false feedback on an alleged personality test. Subsequently, subjects were led to believe that their attractiveness to a physically attractive female student would be compared with that of a male stimulus person. The perceived ability of the male stimulus person to be attractive to females was varied and subjects were given an opportunity to indicate their liking for the male target either before learning the outcome of the female's comparative evaluation or after learning that the female had indicated a preference for the stimulus person. From an analysis of self-esteem threat based on Festinger's theory of social comparison processes (1954), a three way interaction was predicted. Under conditions where subjects had received negative comparison feedback it was predicted that low self-esteem subjects would indicate greater liking for the stimulus person than high self-esteem subjects, regardless of the stimulus person's perceived ability. In contrast, under conditions where comparative evaluation feedback was anticipated, it was predicted that low self-esteem, relative to high self-esteem, subjects, would indicate greater liking for the target perceived to have high ability, but would indicate less liking for the stimulus person perceived to have low ability. The results supported these predictions and are discussed in with respect to furthering our understanding the self-esteem construct and the process of self-esteem maintenance, and the extension of the applicability of social comparison principles.