Failure analysis of structures, a component of the design process, attempts to minimize the costs of mechanical failure by recommending changes in the geometry, materials, processing, or servicing of critical items in order to increase structural reliability. This paper presents a model of failure analysis reasoning and demonstrates techniques of artificial intelligence (AI) that use this model to solve difficult problems of data interpretation. Experiences with a rule-based expert system are used to illustrate problems of knowledge organization and management as the domain of the program gets very large. © 1986.