Previous studies of forest-dwelling bats have identified physical features of trees and forests that correlate with the presence of bats by comparing roost sites to paired, randomly selected sites. This method may be limited if the absence of bats from random sites cannot be confirmed. Our purpose was to address roosting ecology of female big brown bats (Eptesicus fuscus) using a different approach. We quantified relative use of trees with 3 different types of cavity openings (long crevices, multiple holes, or single holes) and compared the relative use of these potential roosts to the availability of each roost type in the study area. Bats used trees with multiple holes and crevices significantly more often than expected based on their availability and trees with single holes less often than expected. Crevice roosts had significantly larger cavities than did single holes and roosting-group size was positively correlated with cavity volume. No relationship was found between cavity volume and tree height or stem diameter of roost trees, 2 variables that have been reported to correlate with roost selection in other studies of forest bats. Examination of our data suggests that the volume of roost cavities may be an important selection criterion for colonial, forest-living bats and that standard interpretations of the roost versus random-tree approach may not accurately identify patterns of roost selection in some systems.