Emerging infectious diseases threaten natural populations, and data-driven modeling is critical for predicting population dynamics. Despite the importance of integrating ecology and evolution in models of host-pathogen dynamics, there are few wild populations for which long-term ecological datasets have been coupled with genome-scale data. Tasmanian devil (Sarcophilus harrisii ) populations have declined range-wide due to devil facial tumor disease (DFTD), a fatal transmissible cancer. Although early ecological models predicted imminent devil extinction, diseased devil populations persist at low densities, and recent ecological models predict long-term devil persistence. Substantial evidence supports evolution of both devils and DFTD, suggesting coevolution may also influence continued devil persistence. Thus, we developed an individual-based, eco-evolutionary model of devil-DFTD coevolution parameterized with nearly two decades of devil demography, DFTD epidemiology, and genome-wide association studies. We characterized potential devil-DFTD coevolutionary outcomes and predicted the effects of coevolution on devil persistence and devil-DFTD coexistence. We found a high probability of devil persistence over 50 devil generations (100 years) and a higher likelihood of devil-DFTD coexistence, with greater devil recovery, than predicted by previous ecological models. These novel results add to growing evidence for long-term devil persistence and highlight the importance of eco-evolutionary modeling for emerging infectious diseases.