Authors: Maureen C Kennedy; Morris C JohnsonFuel treatments are designed primarily to reduce fire hazard and to increase forest resilience. Usually studies of treatment effectiveness compare stand-level variables between treated and untreated forest, with treatments grouped into coarse classifications (e.g., thin-only, thin + prescribed fire, prescribed fire). Fuel treatments are increasingly designed for multiple objectives including wildlife habitat and ecological restoration, resulting in novel treatment prescriptions that do not conform to standard classifications. It is important to understand how various post-treatment vegetation structures and the context of surrounding forest relate to wildfire severity in the treatment unit. We give examples from three fires in the Western US to show that the spatial pattern of severity as the fire moves through the treatment depends on vegetation structure both inside the treatment and in the forest neighboring the fuel treatment. An important predictor of within-treatment severity is mean tree size, which depends on the distribution of tree sizes in the untreated forest. We use our novel methods and results to motivate future studies of fuel treatment design that move beyond simple classifications and binary comparisons of treated and untreated forest to include spatial effects of both fire spread and fire effects, using both simulation and statistical methods.