We simulated population dynamics of bighorn sheep (Ovis canadensis) inhabiting six discrete habitat patches in the Badlands ecosystem, South Dakota. Modeled populations were subjected to a range of potential management actions and rates of disease-causing infection. Simulated disease varied in severity from mild (∼12% mortality) to severe (∼67% mortality), with infections imposed once, at regular intervals, or with a fixed probability each year. In the absence of disease, 200-year extinction rates were uniformly low and insensitive to changes in colonization rate or area of suitable habitat. A single infection, accompanied by change in the area of suitable habitat or colonization rate, resulted in extinction rates of up to 40%, and large changes in average population size (up to 10-fold with changes in area; 4-fold with changes in colonization rate). Simulations with multiple infections, which are probably most realistic, generally resulted in extinction rates that exceeded 20% over a 200-year period. Model results clearly showed that efforts directed toward reducing the frequency or severity of disease are of highest priority for improving the success of attempts to restore bighorn sheep populations. Increases in areas of suitable habitat or improvements to corridors between existing habitat patches were far less likely to improve persistence of simulated sheep populations than reductions in the impact of disease. Although theory predicts that enhanced movements may exacerbate effects of disease, increased colonization rates resulted in relatively small but consistent increases in persistence and average population size for all combinations of parameters we examined.