Combining model predictive control with dynamic programming, a realtime online receding horizon optimal control strategy for medium hybrid electric vehicles is proposed, which based on the driving states of vehicles built by GPS/GIS on board in the future predictive route. The problem of how to reduce the dynamic programming computation and the system variable quantization are studied. The simulation model of predictive control for the fuel economy of the medium hybrid electric vehicles is built. It is verified by the simulation combining C code with MTALAB\\Simulink, that the predictive control algorithm could meet the need of the realtime control of hybrid electric vehicles, and the fuel economy is increased remarkable compared with the foundational vehicles.