Abstract: In order to ensure that the multi-stack fuel cell system (MFCS) can still operate at the maximum efficiency point under the varying load. This paper presented a method of tracking and controlling the maximum efficiency point of MFCS based on the extended kalman filter online identification algorithm. This method utilized the real-time online fitting capability of EKF to quickly realize the identification of the MFCS efficiency/power curve, so as to estimate the maximum efficiency point power of the system in real time. With the power distribution method, the reasonable distribution of the output among the various stacks can be realized to achieve the purpose of maintaining the stable operation of the system at the maximum efficiency point. Finally, the RT-LAB semi-physical hardware-in-the-loop test platform was built and compared with the disturbance observation (P & O) algorithm. The experimental results proved that the method proposed in this paper could quickly estimate the power at the maximum efficiency point of MFCS in real time and realize tracking control. It also reduces the degradation parameters such as the rate of change of fuel cell power and improves the durability of the fuel cell.