We used simulation modeling to analyze wildfire exposure to social and ecological values on a 0.6 million ha national forest in central Oregon, USA. We simulated 50,000 wildfires that replicated recent fire events in the area and generated detailed maps of burn probability (BP) and fire intensity distributions. We also recorded the ignition locations and size of each simulated fire and used these outputs to construct a fire source–sink ratio as the ratio of fire size to burn probability. Fire behavior was summarized for federal land management designations, including biological conservation reserves, recreational sites, managed forest, and wildland urban interface. Burn probability from the simulations ranged from 0.00001 to 0.026 within the study area (mean = 0.0023), and exhibited substantial variation among and within land designations. Simulated fire behavior was broadly related to gradients in fire regimes, although the combined effects of fuel, topography, and simulated weather resulted in fine scale patterns not reflected in ecological and vegetation data. Average BP for the northern spotted owl (Strix occidentalis caurina) nesting sites ranged from 0.0002 to 0.04. Among the 130 different wildland urban interface areas, average BP varied from 0.0001 to 0.02. Spatial variation in the source–sink ratio was pronounced, and strongly affected by the continuity and arrangement of surface and canopy fuel. We discuss the management implications in terms of prioritizing fuel management activities and designing conservation strategies on fire prone landscapes within the 177 million ha national forest network.