Pathogen-driven declines in animal populations are increasingly regarded as a major conservation issue. The Tasmanian devil (Sarcophilus harrisii) is threatened with extinction by devil facial tumor disease, a unique transmissible cancer. The disease is transmitted through direct transfer of tumor cells, which is possible because the genetic diversity of Tasmanian devils is low, particularly in the major histocompatibility complex genes of the immune system. The far northwest of Tasmania now holds the last remaining disease-free wild devil populations. The recent discovery of unique major histocompatibility complex genotypes in the northwestern region of Tasmania has raised the possibility that some animals may be resilient to the disease. We examined the differences in the epidemiology and population effects of devil facial tumor disease at 3 well-studied affected sites in eastern Tasmania and 1 in western Tasmania (West Pencil Pine). In contrast to the 3 eastern sites, there has been no rapid increase in disease prevalence or evidence of population decline at West Pencil Pine. Moreover, this is the only onsite at which the population age structure has remained unaltered 4 years after the first detection of disease. The most plausible explanations for the substantial differences in population effects and epidemiology of the disease between eastern and western sites are geographic differences in genotypes or phenotypes of devils and functional differences between tumor strains in the 2 regions. We suggest that conservation efforts focus on identifying whether either or both these explanations are correct and then, if resistance alleles exist, to attempt to spread the resistant alleles into affected populations. Such assisted selection has rarely been attempted for the management of wildlife diseases, but it may be widely applicable.