The use of composites in high temperature, long lifetime applications requires a basic understanding of composite degradation mechanisms, advances in analytical capabilities, and accurate accelerated and scaled tests. To advance all of these goals, models are proposed based on a variety of fundamental material mechanisms. Thermal, oxygen, and moisture diffusion, chemical reactions, composite micromechanics; modified laminated plate theory, and future mechanics based damage models are used. All models attempt to stay as simple and fundamental as possible. All are coupled, so that interactions between various effects are modeled implicitly. Ongoing efforts at MIT are reviewed here, with some reference to other work, but no attempt is made to do a comprehensive review. Mechanism based models are yielding understanding of the mechanisms behind observed degradation phenomena, are helping to design accelerated tests, and are the first steps towards a predictive capability.