This paper considers a one-product, one-machine production/inventory probelm. Demand requests for the product are governed by a Poisson process with demand size being an exponential random variable. The production facility may be in production or idle; while in production, the facility produces continuously at a constant rate. The objective is to minimize system costs consisting of setup costs, inventory holding costs, and backorder costs. Given a two-critical-number policy, the problem is analyzed as a constrained Markov process using the compensation method. The policy space may then be searched to find the optimal policy.Research supported, in part, by the Office of Naval Research under Contract N00014-75-C-0556.