In the period following wolf (Canis lupus) reintroduction to Yellowstone National Park (1995–2004), the northern Yellowstone elk (Cervus elaphus) herd declined from ∼17 000 to ∼8000 elk (8.1% yr−1). The extent to which wolf predation contributed to this decline is not obvious because the influence of other factors (human harvest and lower than average annual rainfall) on elk dynamics has not been quantified. To assess the contribution of wolf predation to this elk decline, we built and assessed models based on elk-related data prior to wolf reintroduction (1961 to 1995). We then used the best of these models to predict how elk dynamics might have been realized after wolf reintroduction (1995 to 2004) had wolves never been reintroduced. The best performing model predicted 64% of the variance in growth rate and included elk abundance, harvest rate, annual snowfall, and annual precipitation as predictor variables. The best performing models also suggest that harvest may be super-additive. That is, for every one percent increase in harvest rate, elk population growth rate declines by more than one percent. Harvest rate also accounted for ∼47% of the observed variation in elk growth rate. According to the best-performing model, which accounts for harvest rate and climate, the elk population would have been expected to decline by 7.9% per year, on average, between 1995 and 2004. Within the limits of uncertainty, which are not trivial, climate and harvest rate are justified explanations for most of the observed elk decline. To the extent that this is true, we suggest that between 1995 and 2004 wolf predation was primarily compensatory.