Sex allocation in social insects has become a general model in tests of inclusive fitness theory sex-ratio theory, and parent-offspring conflict. Several studies have shown that colony sex ratios are often bimodally distributed, with some colonies producing mainly females and others mainly males. Sex specialization may result from workers assessing their relatedness to male brood versus female brood, relative to the average worker relatedness asymmetry in other colonies of their population. Workers then adjust the sex ratio in their own interest. This hypothesis assumes that workers can recognize the sex of the brood in their colony and selectively eliminate males. We compared the primary sex ratio (at the egg stage) and secondary sex ratio (reproductive pupae and adults) of colonies in the ant Pheidole pallidula. There was a strong bimodal distribution of secondary sex ratios, with most colonies producing mainly reproductives of one sex. In contrast, there was no evidence of a bimodal distribution of primary sex ratios. The proportion of haploid eggs produced by queens was 0.35 in early spring and decreased to about 0.1 in summer. Male eggs also were present in virtually all field colonies sampled in July, although eggs laid at this time of year never give rise to males. All male brood is, therefore, selectively eliminated beginning in July and continue to be eliminated through the rest of the year. Finally the population sex-ratio investment was female-biased. Together, these results are consistent with the hypothesis that workers control the secondary sex ratio by selectively eliminating male brood in about half the colonies, perhaps those with high relatedness asymmetry.