Social insect workers often sacrifice their own reproduction so that they may help relatives to reproduce more successfully. However, genetically dissimilar colony-mates may prefer to aid different kin, and this may result in disruptive conflicts. I developed polymorphic microsatellite genetic loci for the African swarm-founding wasp Polybioides tabidus to examine mechanisms to reduce such conflicts. Swarm-founding wasps have many reproductive queens in their colonies which should lower relatedness, increasing the potential for conflicts. I found that even though P. tabidus colonies contained many queens, relatedness was elevated because new queens were only produced after the number of old queens had been reduced to one, or nearly one. Queens were thus highly related, elevating relatedness in the colony as a whole, and promoting sociality. This unique pattern of queen production is consistent with a worker manipulation of the sex ratios known as cyclical oligogyny. Under cyclical oligogyny, new queens are produced when the colony has few queens, while males are produced when the colony has many queens. The males were indeed produced when queen number was higher, and I found evidence that the workers collectively controlled male production. Colonies which produced normal haploid males also produced diploid males, which have a diploid genome but are homozygous at the sex determining locus. P. tabidus does not appear to effectively distinguish between diploid and haploid males, and diploid males should have occurred in colonies without haploid males too. Their absence indicates that they were actively eliminated from colonies in which the workers did not favor male production. Workers also may have controlled who produced the males. Each worker should prefer to produce the males herself. However, I found that the queens produced the males. This may be explained by collective worker policing because the workers would be more highly related to queen-derived males than to the sons of other workers and should prevent reproduction by other workers. Alternatively, each worker might restrain herself if worker reproduction was costly to the success of the colony. The results of this study indicate that collective worker control is an important mechanism stabilizing cooperation.