Polyurethans are well suited to the design of polymers for good ablation and flame-resistance. Best results in flame resistance may be expected when a composite structure is formed from an inorganic component and a self-extinguishing organic binder. The inorganic component should be present in large amounts so that the space between inorganic particles is small. For low overall density the inorganic material preferably should be foamed. The organic binder should be designed to give minimum flame spread, heat of combustion and smoke generation. The authors believe that best practical results will be achieved if the polymer gives a high melt viscosity under decomposition conditions. Those tests which rate a material as self-extinguishing because it melts and flows away from the flame are considered to be of minimum value, and offer a poor guide to the design of really good flame-resistant polymers as far as actual use is concerned. For minimum flame spread a system should be chosen that will have as little combustible gas evolved as possible, and should include a component which will provide a retarder or inhibitor for free radical decomposition mechanisms. The lowest heat of combustion will usually be favored by the incorporation of components having low hydrogen-to-carbon and low (hydrogen + carbon)-to-oxygen ratios (e.g., urethan, aromatic), and which will decompose by reactions which are not highly exothermic. While most decompositions in an oxygen atmosphere are exothermic, the dissociation of urethan groups may be assumed to be endothermic, and to this extent is favorable. The addition of an effective Lewis acid catalyst to promote carbonaceous char and reduce flammable gas formation will reduce the flame spread, and also lower the heat of combustion. Maximum char formation is desirable to serve as an insulating layer, protecting the polymer below. In addition to the Lewis acid catalyst (as may be derived from phosphorus or many of its compounds, antimony oxide, and similar acid or metal compounds) cyclic structures, especially aromatic, are favorable to char formation. These preferably should be polycyclic to minimize their chance for vaporization. Smoke generation is a function of many things, especially of the rate of burning. Hence the ideas discussed above will also be helpful in this respect. Unfortunately, some of the best structures with regard to flame spread and heat of combustion (acetylenic, aromatic) are not good from the standpoint of smoke generation. This may be because of incomplete combustion of the carbon resulting from relatively low heat of combustion. The best approach appears to be to reduce the rate of flame spread, and the heat of combustion, combined with incorporation of inorganic fillers, so that a low rate of smoke generation follows. This was achieved in Einhorn's foam-glass pellet composite of Table V.