Two experiments were conducted to examine the general hypothesis that a person’s behavior in a social situation is a function of the kind of impression he wants to make on the other person, and that the kind of impression that he wants to make is influenced by the characteristics of the other person. Two experiments were conducted in which male subjects anticipated interaction with either an attractive female (attractive condition), an unattractive female (unattractive condition), or a female whose attractiveness was unknown (control). Each subject first chose either a high, medium, or low status role to enact during the interaction with the female. He then indicated the role which most closely corresponded to his true role. In Experiment I, role choices were significantly biased in the direction of higher status roles in the attractive condition but not in the unattractive or control conditions. These findings were partially replicated in Experiment II, but difference was shown to be a consequence of the fact that the experimenter was an attractive female. The subjects biased their reports of true role status in the direction of higher status, thus leaving little room for change in the selection of role to be enacted during the social interaction. These findings are consistent with the general hypothesis and with exchange theory predictions.