AbstractDemand for some items can depend on the inventory level on display, a phenomenon often exploited by marketing researchers and practitioners. The implications of this phenomenon have received scant attention in the context of periodic‐review inventory control models. We develop an approach to model periodic‐review production/inventory problems where the demand in any period depends randomly, in a very general form, on the starting inventory level. We first obtain a complete analytical solution for a single‐period model. We then investigate two multiperiod models, one with lost sales and the other with backlogging, whose optimal policies turn out to be myopic. Some extensions are also discussed. © 1994 John Wiley&Son