Effective species conservation depends upon correctly identifying the threats that cause decline or hinder recovery. Because estimates of the relative viability of different populations of Endangered African wild dogs Lycaon pictus are most strongly influenced by adult and pup mortality, we analysed rates and causes of mortality in eight wild dog populations under study in southern and eastern Africa. The probabilities of detecting wild dog deaths were influenced by the monitoring methods used. The least biased estimates of mortality causes were obtained through intensive monitoring of radio-collared individuals; this is impossible for pups, however. Mortality patterns varied substantially between populations. Rates of human-caused mortality were higher for wild dogs radio-collared outside protected areas than for those collared inside, but rates of natural mortality were comparable, suggesting that anthropogenic mortality is additive to natural mortality. The relative importance of factors such as snaring and infectious disease also varied regionally. Hence, although our analyses identified no new threats beyond those highlighted in a 1997 range-wide Action Plan, they suggest that local plans will be valuable to target conservation activities more precisely.