Services are often provided by groups. The question of remuneration arises both at the group level and for each individual group member. We examine the question of how relative pay should be designed within the group if all group members are to regard the payment scheme as fair. We use a three-step laboratory experiment to compare which fairness norms are chosen by high-performing and low-performing group members. It turns out that both types of group members prefer the performance pay principle. Support for equal pay is negligible. However, the low performers use their bargaining power to improve their position, but without deviating from the performance principle substantially. A random influence on the performance of the players does not change the results.