Automatic reordering has been an old grudge in supermarket chain stores. It may be accomplished by adequate inventory control model and accurate sales forecasting. But there exists a substantial gap between theory and practice in inventory management. Moreover store personnels have difficulties in sales forecasting under vaious constraints, especially when considering price demand interaction and other irregular factors such as advertising, in-store display, competitor response, calendar variation and weather. Because merchandise reordering is performed on a regular day-to-day basis, we need timely solution than an "optimal," usually time consuming solution. In this regard heuristic methods are preferrable. Electronic-Point-of-Sale (EPOS) allows maintainning perpetual inventories on each item based on timely movement information. Therefore, the experts`` knowledge and data from EPOS can be incorporated into building automatic store reorder system. In this thesis an architecture of knowledge-based store inventory control system for supermarket chain, ISICS (Intelligent Store Inventory Control System), is proposed. We classify merchandise based on the merchandise classification scheme, markup and turnover. We also classify stores based on the demographic characteristics of market area and competition level. Thus merchandise characteristics and store characteristics can be reflected in knowledge base. Matching in ISICS follows top-down approach using hierarchy of such merchandise and store classification. A prototype ISICS has been developed in the UNIK (Unified Knowledge) and LISP environment on IBM PC.