Aiming at the problem of long-term precision navigation of traditional MEMS/GNSS integrated navigation in the case of bad satellite signal, a MEMS/GNSS integrated navigation method based on gray Markov prediction is proposed in this paper. The stability of the system is improved by improving the gray prediction, adding Markov correction links, predicting the GNSS measurement value when the satellite signal is poor, and then replacing the original measurement value, and the result is subject to robust extended Kalman filtering(EKF) to overcome the influence of noise interference. The simulation and sports car experiments verify that the integrated navigation method can still output high-precision navigation results when the satellite signal is poor, and can better overcome the influence of abnormal observations on the system.