We experimentally investigate how reputational concerns affect behavior in repeated Tullock contests by comparing expenditures of participants interacting in fixed groups with the expenditures of participants interacting with randomly changing opponents. When participants receive full information about the choices and earnings of all contestants at the end of each contest we find no difference between contest expenditures in fixed and randomly changing groups. However, when participants only observe their own earnings at the end of each contest they are significantly more aggressive when they interact in fixed groups. This result can be explained by a dominance or status seeking motive.