A comprehensive methodology has been developed for obtaining and using smoke toxicity data for fire hazard analysis. This bench-scale method can simulate diverse fire conditions and identify extremely toxic smoke under both pre- and post-flashover conditions. However, incidence data show that most of the fire deaths in the U.S. occur outside the room of fire origin from smoke and toxic gases that are generated from a fire under post-flashover conditions. Therefore, the most relevant real-scale combustion conditions to simulate in the bench-scale apparatus would be the post-flashover conditions which are achieved by using radiant heat, a high heat flux, and correcting the bench-scale carbon monoxide (CO) results to agree with CO yields observed in real-scale post-flashover fires. The number of test animals (Fischer 344 male rats) is minimized by using the N-Gas Model to estimate the LC 50 value from the chemical analysis of the smoke. The current N-Gas Model predicts the toxicity of complex fire gas mixtures based on a large data base of experimental results of individual and mixed gases that include CO, CO2, reduced O2, HCN, HCl, HBr, and NO X. The prediction is checked with a small number of animal tests and an approximate LC 50 value is determined. The bench-scale results have been validated with full-scale room wall burns of a limited number of materials of widely differing characteristics chosen to challenge the system. The toxic potency values are assessed to determine if the smoke from a material or product is unusually or extremely toxic and can then be used in computations of fire hazard.