We used a broad-scale model based on observations of grizzly bears (Ursus arctos) or their sign, calibrated to reported putative death rate, to appraise current habitat conditions in the Cabinet–Yaak region of Montana. Habitat capability (i.e., potential grizzly bear densities) and regional human population sizes had the greatest effects in this model. We predicted the effects of (1) human population increases ( 150% anticipated by 2023), (2) changes in lethality of humans (i.e., the probability that a human would kill a bear given an encounter), and (3) differences in the ratio of unknown to known bear deaths on the extent and location of potential source areas. We predicted densities of 1.0 and 2.1 grizzly bears/100 km² with and without human impacts, respectively. Under our baseline scenario (3% sustainable mortality and 1:1 ratio of unknown to known bear deaths), we predicted that 2 source areas totaling 9,156 km² and potentially supporting 123 bears occurred in our study area. With projected human population increases, potential source areas and bear populations declined by 33% and 45%, respectively. A spatially uniform increase of 1% in annual death rate (as a surrogate for increased human lethality) reduced potential source areas and bear numbers by 41% and 36%. Source areas and bear numbers declined by 39% and 34% if the ratio of unknown to known grizzly bear deaths was 2:1 versus 1:1. We obtained the best match with current population estimates (about 35 bears) assuming a 2:1 ratio of unknown to known deaths and a very low sustainable death rate of 2%. This implies either high levels of illegal human-caused mortality and low birth and recruitment rates or a population smaller than currently estimated. We conclude that human numbers and human lethality will likely govern the fate of grizzly bears in this region.