Maternally transmitted bacteria that kill male hosts early in their development are found in many insects. These parasites typically infect 1-30% of wild females, but in a few species of insects, prevalences exceed 95%. We investigated one such case in the butterfly Acraea encedon, which is infected with a male-killing Wolbachia bacterium. We measured three key parameters that affect the prevalence of the parasite: transmission efficiency, rate of survival of infected males, and the direct cost of infection. We observed that all wild females transmit the bacterium to all their offspring and that all infected males die in wild populations. We were unable to detect any physiological cost to infection in lab culture. These observations explain the high prevalence of the A. encedon male killer, as theory predicts that under these conditions the parasite will spread to fixation. This will occur provided the death of males provides some benefit to the surviving infected females. The problem therefore becomes why the bacterium has not reached fixation and driven the butterfly extinct due to the shortage of males. We therefore investigated whether males choose to mate with uninfected rather than infected females, as this would prevent the bacterium from reaching fixation. We tested this hypothesis in the "lekking swarms" of virgin females found in the most female-biased populations, and were unable to detect any evidence of mate choice. In conclusion, this male killer has spread to high prevalence because it has a high transmission efficiency and low cost, but the factors maintaining uninfected females in the population remain unknown.