Studies of social power use have frequently employed self-report instruments that are prone to response bias. In the present study, an experimental design was conducted in which 100 participants were asked to gain compliance from in-group and out-group members. We tested whether harsh tactics—often used for elevating self-image at the dyadic level—also provide a means for gaining advantage at the group level. For this purpose, self-esteem and self-efficacy were examined as possible moderators. Findings indicated that self-esteem interacted with target group: Low self-esteem participants used harsh tactics more frequently toward in-group than out-group members; and moderate and high self-esteem participants used harsh tactics more frequently toward out-group than in-group members. The process involved in this interaction is discussed.