Split–sex–ratio theory assumes that conflict over whether to produce predominately males or female reproductives (gynes) is won by the workers in haplodiploid insect societies and the outcome is determined by colony kin structure. Tests of the theory have the potential to provide support for kin–selection theory and evidence of social conflict. We use natural variation in kinship among polygynous (multiple–queen) colonies of the ant Formica exsecta to study the associations between sex ratios and the relatedness of workers to female versus male brood (relatedness asymmetry). The population showed split sex ratios with about 89% of the colonies producing only males, resulting in an extremely male–biased investment ratio in the population. We make two important points with our data. First, we show that queen number may affect sex ratio independently of relatedness asymmetry. Colonies producing only males had greater genetic effective queen number but did not have greater relatedness asymmetry from the perspective of the adult workers that rear the brood. This lack of a difference in relatedness asymmetry between colonies producing females and those producing only males was associated with a generally low relatedness between workers and brood. Second, studies that suggest support for the relatedness–asymmetry hypothesis based on indirect measures of relatedness asymmetry (e.g. queen number estimated from relatedness data taken from the brood only) should be considered with caution. We propose a new hypothesis that explains split sex ratios in polygynous social insects based on the value of producing replacement queens.