In this article, we report the use of a variety of analytical methods, in particular, solid-state H-1-NMR and C-13-NMR to characterize the relationship between the condensed-phase chemistry and burning behavior as determined by a series of combustion tests for two commercially derived flexible polyurethane foams, one combustion-modified. The combustion tests showed that the foams met several regulatory requirements in terms of their fire performance, whether or not they were combustion-modified. Both foams passed the MV SS 302 and CAL 117 small-flame tests. The nonmodified foam failed the Crib 5 test, but this test had a much larger ignition source. The particular problem with the nonmodified foam was melt drip into the flame zone. This led to a steady maintenance of the fuel feed and a rapid escalation of the fire. In contrast, the combustion-modified foam showed little melt drip and self-extinguished. Thermal analysis data for the two foams showed that melamine acted in part as an endothermic heat sink This alone did not account for the much reduced melt flow and drip of the combustion-modified foam, but the solid-state H-1-NMR data clearly showed that the molecular mobility of the combustion char from combustion-modified foam was lower than the unmodified foam char, which indicated that the flame-retardant formulation in the combustion-modified foam acted by a condensed-phase mechanism. (c) 2006 Wiley Periodicals, Inc.