In mountainous areas with sufficient snowfall, avalanche chutes are an important component of grizzly bear (Ursus arctos) habitat. Therefore, regional land-use plans have recommended retaining adjacent forest buffers to maintain security and thus reduce potential impacts of clearcut forest harvesting. Our objective was to determine if forest buffers affected selection of avalanche chutes by grizzly bears, while accounting for factors such as vegetation composition and other physical attributes. We used radio-location data from 61 grizzly bears collected between 1994 and 2000 in southern British Columbia, mapped a sample of avalanche chutes (1,045), and quantified the amount of forb, shrub, tree, and non-vegetated cover within each chute. We also measured forested buffer width on each side of the chute, solar radiation, chute size, chute frequency (no. of chutes/km), and the area of clearcut logging adjacent to chutes. Each avalanche chute was the sample unit and the number of grizzly bear radiolocations was the dependent variable. We found that natural biophysical attributes were the strongest factors predicting the level of avalanche chute use by bears. Frequency of large chutes (>100 m wide), chute area, forb content, and solar radiation all positively affected use by bears. Larger avalanche chutes had a higher proportion of forb cover than smaller chutes, and more of these large chutes per unit area provided increased forage opportunities. Based on multivariate analyses, forested buffer width or the amount of clearcut logging were not strong factors predicting the level of use. However, a post hoc univariate analysis revealed that clearcut logging reduced the amount of bear use of the best avalanche chutes (large and abundant chutes). Furthermore, because a portion of our study area contained logging but no vehicle traffic, we concluded that it was the removal of tree cover, rather than displacement by vehicles, that caused the observed pattern. Although our multivariate models did not perform well using independent validation in a different geographic area, 4 factors were consistently important (large and abundant chutes, forb content, with a negative but weaker influence of clearcutting), suggesting broad applicability of these factors in mountainous ecosystems.