The present experiment introduces a modification of the iterated prisoner's dilemma (PD). In contrast to classical dilemma situations with only one interaction partner, participants (N = 120) interacted with five fictitious interaction partners within one game, either in a random order (change condition) or against each of the interaction partners in succession (block condition). The authors assume that the change condition simulates the social interactions of a real environment more accurately and that individual memory skills are more important in the change condition as compared to the block condition. As dependent variables, the participants' score in the game was recorded, as well as the participants' memory performance concerning information about their interaction partners. Results show that good memory performance with respect to biographical information leads to higher scores only in the condition with changing interaction partners, but not in the block condition.