Habitat suitability models are useful tools for investigating species’ requirements and forming a resistance surface for predicted connectivity. Robustly predicting habitat and corridors are salient for species that are rare, declining, dispersal limited, or impacted by emerging threats. We evaluated predicted habitat, connectivity, and fire risk for Pacific martens (Martes caurina) in northern California with a large, long-term, dataset (n = 32 GPS-collared martens). We used MaxEnt to model spatially-thinned locations and paired variables optimized for spatial scale. Our top model included elevation, tree biomass, slope, tree density, and canopy cover standard deviation. Independent GPS and camera data suggested predictive accuracy. We created a resistance surface using a negative exponential function. We identified 8 core areas and 10 linkages. Across our modeled area, 24 fires burned 2000-2020. We predicted 41% of the modeled core and corridor network at moderate wildfire hazard potential. Following the completion of our evaluation, the 2021 Dixie fire burned >38% of our core and corridor network, >20% at moderate or high severity. We reflect on opportunities following unexpected disturbances.