We fit data on elk (Cervus elaphus) population size and composition, survival rates measured from their first week of life, reported harvest, and local weather to a series of alternative population models of the elk herd in Jackson, Wyoming, USA, for the period 1980–2002. Data were corrected for biases in aerial survey visibility, misclassification of juveniles in ground surveys, and harvest reporting. The models included explanatory variables for sex, age, population size, weather, and autocorrelation of survival rates in different periods. Using information–theoretic model selection, we identified the most strongly supported models and effects. Model complexity ranged from 12 to 70 fitted parameters, and the best-supported model contained 25 parameters. We estimated annual natural survival (excluding harvest) of mature (≥1 yr) elk of 96.8% (SE = 1.5%) for males and 97.2% (SE = 2.2%) for females. Natality was 60.4 juveniles/100 mature females (SE = 3.9 juveniles/100 mature females). Sex ratio at birth strongly favored females (45.8% males, SE = 1.6%, Akaike weight = 99.9%). The dynamics of this population were well explained by annual variation in survival of neonates (birth to 31 Jul), juvenile survival during late winter (20 Feb–19 May), and harvest. Survival of neonates was correlated with several weather covariates that apparently affected nutritional status of their mothers. Survival of juveniles during late winter was related to weather conditions during the preceding summer and early winter. We found a compensatory effect of juvenile harvest on subsequent juvenile survival in late winter; 89% of increased juvenile harvest was offset by reduced natural mortality. We also found evidence for a decline in survival of neonates with increasing population size (density dependence). However, the density effect was weak at current population size and recent supplemental feeding rates. Thus, only continued or increased female harvest can maintain this population at current or lower levels if current feeding policies are continued—unless disease prevalence, predator impacts, or other factors substantially alter the historical dynamics. Simulations suggested that harvest rates of mature females must be increased to 15.1% from recent levels of 11.9% to reduce the current population of 15,680 elk (SE = 407) to the target population size of 11,029 set by the Wyoming Game and Fish Department (WGFD). Sensitivity of equilibrium population size at the WGFD target level to harvest rate was very high, requiring regular monitoring and adjustment of harvest to maintain a stable population.