Although understanding natural mortality rates of ungulate populations is essential for effective management, published data on adult survival from unharvested pronghorn (Antilocapra americana) populations in the Northern Great Plains is limited. We estimated seasonal adult survival rates and cause-specific mortality of an unharvested pronghorn population in Custer State Park, S.D. We assessed the relative importance of sex, age, year, and season in explaining pronghorn survival rates using an information-theoretic approach. We captured and radio-collared 26 male and 24 female adult pronghorn from fall 2005 through spring 2008. We observed higher predation rates and lower survival of adult pronghorn in CSP compared to other populations in the region, but similar to the pronghorn population in Yellowstone National Park. We documented 23 deaths (10 females, 13 males) of the 50 radio-collared pronghorn from Nov. 2005-Nov. 2008. Predation by mountain lions (Puma concolor) and coyotes (Canis latrans) accounted for 69.5% of all mortalities. The season model received the greatest support although there also was strong support for the season × sex model. Seasonal survival for males and females was >0.90 for the winter-grouping and breeding seasons but fell to 0.791(95% CI 0.644-0.887) and 0.837 (95% CI 0.706-0.916) for females and males, respectively, during the small group-parturition season. A dense predator population, as well as a higher vulnerability to predation when pronghorn are solitary or in small groups, may explain the lower survival during these time periods. If population estimates fall below management goals, management actions aimed at reducing predator cover may be beneficial to adult pronghorn. Managers of pronghorn populations near forested and rugged areas and that are sympatric with dense predator populations should consider adult survival may be lower than observed in Great Plains populations.