Abstract: This paper aims to solve the problems of great delivery pressure and high inventory cost faced by the auto parts manufacturer, which is under traditional just-in-time production mode and is non-dominant in the supply chain. The auto parts manufacturer who uses make-to-order production mode is taken as an example and the raw materials are classified into three types by two steps. Multi-period dynamic programming models, which minimize the inventory costs for the three types of raw materials are formulated and solved. Finally, through numerical experiments with the real operation data of M company, we find that the inventory turnover period and inventory cost with the optimal purchasing decisions are reduced obviously, which verifies the feasibility and effectiveness of the optimal ordering polices.