Habitat selection by dispersers is the focus of much theoretical models, most of which are based on the assumption of negative density dependence. The archetype of these models is the ideal free distribution, characterized by an evolutionary stable state where more competitors aggregate in better habitats, so that the fitness benefit of resource abundance is equally offset by the cost of competition in all habitats. In this study, we used parentage analysis on microsatellite genotypes to test the ideal free distribution in a natural population of aphid parasitoids. Parentage analysis was conducted on parasitoids emerging from aphid colonies. We inferred the number of foundress females which had reproduced in each colony, as well as the number of offspring for each foundress. As predicted by the ideal free distribution, the number of offspring per foundress per colony did not depend on the number of hosts per colony. However, contrary to ideal free distribution predictions, it was affected by the number of foundresses per colony. In surprising contrast with the basic assumption of negative density dependence, individual fitness increased with the number of foundresses. Moreover, parentage analysis revealed a very low number of offspring per foundress per colony (mean = 1.8). This observed distribution questions the validity of classical models of habitat choice based on competition. Indeed, our results provide a new illustration reinforcing a growing body of theory and data on positive density dependence. Our results also suggest that the avoidance of hyperparasitism and predation, although generally neglected, may shape the distribution of parasitoids in the field.