In most inventory models one of the costs considered is shortage cost. However, this cost is often very hard to measure. For the models considered in this paper, instead of introducing a shortage cost, we define a stock-out penalty function and impose a constraint on it. For example, a constraint on the probability of stock-out can be viewed this way. We exhibit the general, randomized, optimal policy and give graphical and numerical methods of calculating it. We also discuss how our stock-out penalty function is related to shortage cost.