Many firms are exploring production and supply chain strategies when customers may be segmented into different classes based on service level or priority. Such segmentation can result in a more efficient production system as well as a better match between supply and demand. In this research, we analyze a system with customer classes 1 and 2, where customer class 1 has a higher priority of fulfillment than customer class 2 in the same period. We develop an optimal production and inventory strategy that rations current and future limited capacity between customer classes 1 and 2, through reserving inventory for the future and accepting orders now for future delivery when demand and production are general stochastic functions. We show that a modified order-up-to policy ( S*, R-i*, B-i*) is optimal in each period. S* is the targeted inventory level after production at the beginning of the period; R-1* represents the optimal inventory to be protected from being sold to both classes, and R-2* is the additional amount of inventory to protect from class 2. B-2* is the optimal amount of future capacity to make available to both classes through backlogging, and B-1* is the additional backlogging amount for class 1. Computational analysis shows that the differentiation strategy can result in a significant profit improvement over a traditional inventory policy.