We proposed a prototype design of an web-based intelligent inventory management system for logistics management focused on its functional and working algorithm. Intelligent system concepts and export knowledges concerning inventory management are adopted to develop this system composed of three layers such as network communication layer, problem-solving layer and user interface layer. The problem-solving layer is composed of inference engine, model base, knowledge base and data base agent. We will describe the mechanism of the proposed web-based lIMS with an emphasis on the role of each agents. The IIMS can select an appropriate inventory model considering logistics situation and provide the solution for it by utilizing stored algorithm of the selected inventory model.