We used the Fire and Fuels Extension to the Forest Vegetation Simulator (FFE-FVS) to simulate fuel treatment effects on 45,162 stands in low- to midelevation dry forests (e.g., ponderosa pine (Pinus ponderosa Dougl. ex. P. & C. Laws.) and Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) of the western United States. We evaluated treatment effects on predicted post-treatment fire behavior (fire type) and fire hazard (torching index). FFE-FVS predicts that thinning and surface fuel treatments reduced crown fire behavior relative to no treatment; a large proportion of stands were predicted to transition from active crown fire pre-treatment to surface fire post-treatment. Intense thinning treatments (125 and 250 residual trees*ha-1) were predicted to be more effective than light thinning treatments (500 and 750 residual trees*ha-1). Prescribed fire was predicted to be the most effective surface fuel treatment, whereas FFE-FVS predicted no difference between no surface fuel treatment and extraction of fuels. This inability to discriminate the effects of certain fuel treatments illuminates the consequence of a documented limitation in how FFE-FVS incorporates fuel models and we suggest improvements. The concurrence of results from modeling and empirical studies provides quantitative support for 'fire-safe' principles of forest fuel reduction (sensu Agee and Skinner 2005. For. Ecol. Manag. 211: 83-96).