Wildfires have important effects on wildlife habitat. Within fire perimeters, ‘fire refugia’ are unburned or less severely burned areas that contain residual pre-fire habitat structures and facilitate species persistence following fire. However, fire refugia may vary in their relative quality. The ability to evaluate the relative quality of fire refugia for wildlife habitat may help efficiently allocate resources for post-fire conservation planning, especially for at-risk species. Using spatial overlays of habitat models and burn severity data, we developed the refugia index, a flexible multi-scale, multi-criteria fuzzy logic model to assess the value of unburned areas for wildlife habitat affected by wildfires and other disturbances. We conducted a case study of the northern spotted owl (Strix occidentalis caurina) and applied our method to 4278 unburned areas in 15 fires that each burned > 400 ha of spotted owl nesting/roosting habitat in the eastern Cascade Mountains, U.S.A. In our study area, unburned areas represented ~ 1.7% of the area within the fire perimeters. Most fire refugia were low quality, because unburned areas were small relative to the desired nest patch size of spotted owls and contained little suitable nesting/roosting habitat. However, each fire contained a few high-quality fire refugia, because they co-occurred with suitable nesting/roosting habitat and were surrounded by desirable habitat characteristics for spotted owls. Our method illustrates how land managers can use readily available spatial data and analysis tools to understand fire effects on species’ habitat (e.g., habitat lost) and prioritize areas for post-fire conservation and management.