Social insects have become a general model for tests of sex allocation theory. However, despite tremendous interest in the topic, we still know remarkably little about the factors that cause dramatic differences in sex allocation among local populations. A number of studies have suggested that environmental factors may influence sex allocation in ant populations. In polygynous (multiple queens per nest) populations of the ant Formica exsecta, sex allocation is extremely male biased at the population level, with only a small proportion of nests producing any gynes (female reproductive brood). We analysed the proportion of gyne-producing nests in 12 F. exsecta populations during three successive breeding seasons and found considerable temporal and spatial variability in the proportion of gyne-producing nests. The populations differed in a number of characteristics, including elevation, nest density, size of the nest mound, and number of nests per population. However, the proportion of gyne-producing nests was not associated with any of these geographic and demographic variables. Moreover, differences between populations in the production of gynes were not consistent between years. Thus, the proportion of gyne-producing nests appears to vary stochastically, perhaps because of stochastic variations in environmental factors. For example, year-to-year variations in the proportion of gyne-producing nests were associated with differences in spring weather conditions between years. The finding that gyne production varies greatly between years suggests that it may not always be adaptive at a local scale.