The integrated deviation of CNS/INS is an important factor that restricting the accuracy of integrated navigation system. A calibration method of CNS/INS integrated navigation system based on fixed point smoother algorithm is presented in this paper. The deviations parameters can be estimated optimally by fully utilize the measurement information using the smoother algorithm that based on convention Kalman filtering process. Simulation results show that the smoother algorithm can realize more stability and more accuracy estimated result and the validity of this method are verified.