In the context of decision support systems, a model management system (MMS) is analogous to a database management system providing support for the various phases of a modeling life-cycle while insulating user from the physical aspects of model base storage and processing. An underlying premise is the recognition of models as a resource that need to be managed and modeling as an activity that need to be supported within organizations. The recent developments of the Web and distributed computing environments is creating ever increasing demands for sharing and reusing heterogeneous models over corporate intranets and the Internet. To this end, this paper presents a semantic Web services-based architecture for MMS. The architecture is based on service-oriented principles and web services standards to support model sharing and reuse in a distributed environment. Integral to the architecture is the use of ontologies and artificial intelligence (AI) planning techniques to facilitate model composition. We demonstrate the features and advantages of the proposed architecture using representative examples. We conclude with a discussion of the current state of implementation and plans for future work.