The authors examined the impact of power on in-group bias by manipulating group members' power over the in-group and power over the out-group as orthogonal factors. Each factor had 3 levels: 0%, 50%, and 100%. Participants were 216 male pupils (12-13 years old). Participants showed no in-group bias when they had 0% control over the in-group, strong in-group bias with 50% control, but less in-group bias with 100% control. Participants showed more in-group bias when they had 0% control over the out-group than when they had 50% or 100% control. The combination of these 2 main effects resulted in the noblesse oblige effect: Group members with complete control over both in-group and out-group expressed less in-group bias than did group members who shared control with an out-group.