We examined foraging-habitat selection of Black-backed Woodpeckers (Picoides arcticus) in burned forests of southwestern Idaho during 2000 and 2002 (6 and 8 years following wildfire). This woodpecker responds positively to large-scale fire disturbances and may be at risk from logging and post-fire management. With 100 radio-locations of four adult males, we used resource-selection probability functions in logistic form in an information-theoretic framework to model the Black-backed Woodpecker's selection of foraging habitat at fine and coarse spatial scales. Fine-scale data included characteristics of the foraging tree (tree level) and local habitat surrounding foraging trees (plot level, 0.04 ha), whereas coarse-scale data (224–778 ha) included patch characteristics within home ranges. Model selection by Akaike's information criterion identified a multi-scale model containing tree- and plot-level covariates, and their interaction, as the best model to characterize foraging trees. The positive effect of interaction between foraging-tree diameter and plot-level tree density suggested that foraging Black-backed Woodpeckers select both patches with dense trees and larger-diameter trees. Fire severity was not important, possibly because of the time since fire and the birds' habit of foraging on recently dead trees in adjacent unburned forests. Salvage logging that retains large-diameter trees in clumped distributions is most likely to provide long-term foraging habitat for Black-backed Woodpeckers in dry coniferous forests of the interior western U.S. Following wildfire, conservation of forest patches containing weakened trees may extend the suitability of habitat for foraging Black-backed Woodpeckers by up to 8 years.