Male-killing bacteria are cytoplasmic sex-ratio distorters that are transmitted vertically through females of their insect hosts. The killing of male hosts by their bacteria is thought to be an adaptive bacterial trait because it augments the fitness of female hosts carrying clonal relatives of those bacteria. Here we attempt to explain observations of multiple male-killers in natural host populations. First we show that such male-killer polymorphism cannot be explained by a classical model of male-killing. We then show that more complicated models incorporating the evolution of resistance in hosts can explain male-killer polymorphism. However, this is only likely if resistance genes are very costly. We also consider the long-term evolutionary dynamics of male-killers, and show that evolution towards progressively more 'efficient' male-killers can be thwarted by the appearance of host resistance. The presence of a resistance gene can allow a less efficient male-killer to outcompete its rival and hence reverse the trend towards more efficient transmission and reduced metabolic load on the host.