Several hypotheses have been put forward to explain the adaptive significance of interspecific variation in mating frequencies by eusocial hymenopteran queens. Four of these hypotheses assert that polyandry is advantageous to queens because of the resultant increase in genetic variability within colonies (referred to as the ''GV'' hypotheses). Here we compare the frequency of polyandry between monogynous (single-queen) and polygynous (multiple-queen) ant species to test the hypotheses that (I) multiple mating functions primarily to increase intracolonial genetic variability, and (2) mating has costs (such as increased energetic losses or risk of predation or venereal disease). If one of the GV hypotheses is true and mating is costly, the frequency of polyandry should be lower in polygynous species (in which the presence of multiple queens results in low relatedness among workers) than in monogynous species. As predicted by the GV hypotheses, polyandry is less common among polygynous than among monogynous species. Furthermore, it seems that the causal relationship underlying this association is that the number of matings by queens depends on the number of queens present in the colony (rather than the number of queens being influenced by the number of matings), which also supports the GV hypotheses together with the assumption that mating has costs.