Counts of northern Yellowstone elk (Cervus elaphus) in northwestern Wyoming and adjacent Montana, USA, have decreased at an average rate of 6-8% per year since wolves (Canis lupus) were reintroduced in 1995. Population growth rates of elk are typically sensitive to variations in adult female survival; populations that are stable or increasing exhibit high adult female survival. We used survival records for 85 radiocollared adult female elk 1-19 years old to estimate annual survival from March 2000 to February 2004. Weighted average annual survival rates were approximately 0.83 (95% CI = 0.77-0.89) for females 1-15 years old and 0.80 (95% CI = 0.73-0.86) for all females. Our estimates were much lower than the rate of 0.99 observed during 1969-1975 when fewer elk were harvested by hunters, wolves were not present, and other predators were less numerous. Of 33 documented deaths included in our analysis, we attributed 11 to hunter harvest, 14 to predation (10 wolf, 2 unknown, 1 cougar [Puma concolor], and 1 bear [Ursus sp.]), 6 to unknown causes, and 2 to winter-kill. Most deaths occurred from December through March. Estimates of cause-specific annual mortality rates were 0.09 (0.05-0.14) for all predators, 0.08 (0.04-0.13) for hunting, and 0.07 (0.03-0.11) for wolves specifically. Wolf-killed elk were typically older (median = 12 yr) than hunter-killed elk (median = 9 yr, P = 0.03). However, elk that winter outside the park where they were exposed to hunting were also younger (median = 7 yr) than elk that we did not observe outside the park (median = 9 yr, P < 0.01). Consequently, differences in ages of elk killed by wolves and hunters may reflect characteristics of elk exposed to various causes of mortality, as well as differences in susceptibility. Unless survival rates of adult females increase, elk numbers are likely to continue declining. Hunter harvest is the only cause of mortality that is amenable to management at the present time.