An industrial production process is modelled as a discrete-time system with a disturbance input due to sales. The state consists of rates of flow of parts or subassemblies at various work stations, backlogs of parts awaiting processing, and the inventory level of the finished product. The control variables are the man-hours scheduled for various work processes. A quadratic performance criterion is minimized so as to keep state and control variables near desired values. Dynamic programming is used and numerical examples are provided.