Two types of male-killing cytoplasmic gene are distinguished by their time of action. Early male killers strike embryos, whereas late killers, so far only observed in mosquitoes, kill fourth instar larvae. Maternal inheritance of cytoplasmic factors underlies the sex bias in mortality in both instances. However, the evolutionary logic of early and late killing is, it is argued, otherwise different. Early male killing is interpreted in terms of kin selection. The death of male embryos early on allows resources (food or space) to be redistributed to the surviving females. A simple model relying on the redirection of resources after male death indicates that so long as a more than critical proportion of `freed' resources goes to females containing clonal relatives of the male-killing cytoplasmic gene, then that gene can invade and go to an equilibrium. This equilibrium is significant in that it allows a population to avoid extinction prior to the evolution of a nuclear repressor gene. The distribution of male killing is discussed in the light of these findings. The late male-killing microsporidians of mosquitoes differ from early male killing factors in possessing horizontal, as well as vertical, transmission between generations. It is argued that to maximize the efficiency of horizontal transmission it is in the interests of the microsporidians to maximize their number by waiting for the host to grow before killing it. Not all microsporidians kill only the male mosquito, some also kill the female. The variety of mosquito-microsporidian pathologies is interpreted by considering a trade-off between the efficiency of vertical against horizontal transmission. A review of possible incidences of cytoplasmic male killing is presented.