Abstract:  Auto manufacturers provide customers with a basic twodimensional warranty that quantifies vehicle lifetime with two metrics: time and mileage. Under such a warranty policy, it is very important to control the spare part inventory. With sales prediction as its starting point, the number of automobiles under guarantee in some period is determined according to twodimensional warranty policies. It is known that the failure rate of some nonrepairable items is determined by age or usage. Based on this property, a model is proposed to forecast the demands of spare parts. Then, with the demands obtained, a multistage (Q, r) dynamic inventory control policy is presented. Case studies are given to show the effectiveness for inventory cost reduction by using the proposed method.