In this paper, a simple regression model is developed for the estimation of optimal inventory-control parameters for items with lumpy demand patterns when a service level is specified. The control discipline is the (s, S) policy with continuous review. The lead time is assumed to be known and constant, and there is assumed lo be complete backlogging of unfilled demand. A lead-time demand distribution of stuttering Poisson nature is assumed. Customer service is defined as the probability of having a non-negative net inventory at an arbitrary time. A grid of 2688 known optimal policies is used to provide data for calibration of the regression model. Extensive computational results show that for the range of service level considered in this paper, the model developed provides close approximations to the optimal values of s and S for wide ranges of demand and cost parameters