A dynamic model of a direct district heating system (DDHS) is developed, and an energy optimal control strategy is designed. The DDHS is characterized as a system with long transport time delay. The use of a Smith predictor (SP) to deal with this type of time delay is explored. An SP is designed by using the reduced-order dynamic model and implemented on the full-order model. Also, optimal set point profiles of supply water temperature as a function of outdoor air temperature have been determined. The simulation results show that the SP works effectively in disturbance rejection compared with the nominal control design. The use of optimal set point control strategy resulted in energy savings of the order of 19-32% when the influence of internal load was considered. Copyright (C) 2004 John Wiley Sons, Ltd.