A power control strategy was proposed for hybrid electric vehicle (HEV) to improve fuel economy in real time based on covariate adaptation. The principle and goal of covariate adaptation was realization of engine highefficient working, which means the selection of covariate was determined by the high efficiency range of engine. The strategy was realized through three successive steps. First, the future demand energy should be predicted and simplified to the prediction of demand energy level. A fuzzy controller was established to classify the demand energy of vehicle. Then, the high efficiency range was adjusted according to the predicted demand energy level. Meanwhile, the SOC soft constraint range designed was also adjusted. Finally, the covariate was adaptively adjusted according to the principle of high efficiency within the SOC soft constraint range, otherwise it would be adjusted by SOC. The proposed method shows a good control effect in both optimizing power distribution and controlling SOC range on a mild hybrid vehicle simulation platform.