The problem of inventory classification in spare parts inventory system is considered. The classification is performed prior to determining the inventory review and replenishment policies, and these are a basis for developing and applying mathematical models. A good inventory classification has to take into account the total inventory cost per time period rather than usage value which is used in classical ABC classification. The problem arises from the fact that it is very hard to measure total cost, especially stock-out cost which are of particular importance in a maintenance system. Therefore, the classification is to be made on the basis of attributes describing spare parts. Inference mechanism as a part of proposed expert system uses data about spare parts from an existing data base and heuristic rules from a knowledge base.