The main objective of this study is to clarify the impact of energy demand prediction accuracy on optimal operating result of residential energy supply system. In this paper, the energy system consists of polymer electrolyte fuel cell cogeneration system and electrical-driven heating, ventilating and air conditioning system. Primary energy of operational results with various demand prediction patterns are calculated in order to analyze the effect of demand prediction error. As a result, it is concluded that prediction error of energy demand causes 54.3 MJ/day increase of daily primary energy in the worst case, and annual primary energy consumption can be reduced by 13.2% with the best prediction in comparison with the worst one. Furthermore, the impact of energy demand prediction accuracy on daily primary energy has relationship with the amount of hot water demand. Therefore, the effectiveness of using the amount of hot water demand as a parameter for demand prediction is suggested in terms of energy saving.