The purpose of this lecture is to assess the current state of our ability to simulate the consequences of a fire in a large building, and suggest some areas where improvement is needed. Attention is focused on the coupling of fire dynamics simulations and heat transfer analyses to each other and to structural analyses of the damaged building. The role that uncertainty in “input parameters” resulting from coupling a sequence of complex simulations is considered. The methodology used in the NIST investigation into the collapse of the World Trade Center Towers will be described from this perspective. The intent is not to summarize the results of the investigation, but rather to provide a specific context that illustrates the strengths and weaknesses of the methodologies employed. Research needs are emphasized by examination of some basic problems in fire-structure interactions.