Hamilton's concept of local mate competition (LMC) is the standard model to explain female–biased sex ratios in solitary Hymenoptera. In social Hymenoptera, however, LMC has remained controversial, mainly because manipulation of sex allocation by workers in response to relatedness asymmetries is an additional powerful mechanism of female bias. Furthermore, the predominant mating systems in the social insects are thought to make LMC unlikely. Nevertheless, several species exist in which dispersal of males is limited and mating occurs in the nest. Some of these species, such as the ant Cardiocondyla obscurior, have evolved dimorphic males, with one morph being specialized for dispersal and the other for fighting with nest–mate males over access to females. Such life history, combining sociality and alternative reproductive tactics in males, provides a unique opportunity to test the power of LMC as a selective force leading to female–biased sex ratios in social Hymenoptera. We show that, in concordance with LMC predictions, an experimental increase in queen number leads to a shift in sex allocation in favour of non–dispersing males, but does not influence the proportion of disperser males. Furthermore, we can assign this change in sex allocation at the colony level to the queens and rule out worker manipulation.