Kin selection theory predicts that, in social Hymenoptera, the parentage of males should be determined by within-colony relatedness. We present a model showing that, when sex ratios are split (bimodal) as a function of colony kin structure, the predictions of kin selection theory regarding the occurrence of worker reproduction and policing (prevention of worker reproduction) require modification. To test the predictions of kin selection theory and our model, we estimated using microsatellites the frequency of worker-produced male eggs and adults in the facultatively polygynous (multiple-queen) ant Leptothorax acervorum. Analysis of 210 male eggs and 328 adult males from 13 monogynous (single-queen) and nine polygynous colonies demonstrated that the frequency of worker-produced males was low (2.3–4.6% of all males) and did not differ significantly between colony classes or between eggs and adults. This suggested workers’ self-restraint as the cause of infrequent worker reproduction in both colony classes. Such an outcome is not predicted either by comparing relatedness values or by our model. Therefore, it appears that factors other than colony kin structure and sex ratio effects determine the pattern of male parentage in the study population. A likely factor is a colony-level cost of worker reproduction.