The absence of information about how abundance varies across species' ranges restricts most modeling and monitoring of climate change responses to the range edge. We examine spatial variation in abundance across the northeastern range of North American beaver (Castor canadensis), evaluate the extent to which climate and nonclimate variables explain this variation, and use a species–climate envelope model that includes spatial variation in abundance to predict beaver abundance responses to projected climate change. The density of beaver colonies across Québec follows a roughly logistic pattern, with high but variable density across the southern portion of the province, a sharp decline in density at about 49°N, and a long tail of low density extending as far as 58°N. Several climate and nonclimate variables were strong predictors of variation in beaver density, but 97% of the variation explained by nonclimate variables could be accounted for by climate variables. Because of the peak and tail density pattern, beaver climate sensitivity (change in density per unit change in climate) was greatest in the interior and lowest at the edge of the range. Combining our best density–climate models with projections from general circulation models (GCM) predicts a relatively modest expansion of the species' northern range limit by 2055, but density increases in the range interior that far exceed those at the range edge. Thus, some of the most dramatic responses to climate change may be occurring in the core of species' ranges, far away from the edge-of-the-range focus of most current modeling and monitoring efforts.