For the survivability analysis of a weapon system, an analyst has to consider and choose from a large number of available models the set of models that best fits the study requirements. This paper presents a knowledge-based approach for this Model Management System (MMS) problem. A hierarchical structure of the models is used in the MMS, where a lower level model's output must match the input requirements of a higher level model. The search procedure for the appropriate model set is driven by the user defined weapon analysis requirements. This type of search procedure ensures that the final output of the selected models would provide the required information to the user and the set of selected models would be properly matched. A prototype of the knowledge system has been developed using a commercially available PC-based shell.