Since its emergence, devil facial tumour disease (DFTD) has spread throughout most of the geographic range of the Tasmanian devil, causing >90% local population decline, >80% overall decline and cascading effects on the ecosystem. We developed a decision tree to guide research and management of this cancer. The devil–DFTD system gives an opportunity to study a wildlife disease in all stages of existence across the entire geographic range of a natural host. Despite predictions of extinction, devils persist. State–space models show that individuals with higher fitness (the larger, more dominant individuals responsible for most of the biting) are more likely to become infected themselves. Individual-based models in which demographic parameters depend on the size of the tumours carried by individual hosts show that DFTD epidemics operate on a much slower timescale than those of viral or bacterial diseases. Following an initial epidemic peak, the consequences for a general epidemic may be coexistence, even in the absence of evolutionary changes in either host or pathogen. Multiple lines of evidence show that Tasmanian devils are evolving in response to DFTD. Conservation efforts are now shifting from managing for extinction to managing for persistence.