Abstract The attitude combination of celestial navigation/Strapdown inertial(CNS/SINS) integrated navigation system can make the attitude angle converge and restrain the divergence of position and velocity effectively. To improve the accuracy of the integrated system, we design the UKF algorithm of CNS/SINS integrated navigation system. Firstly, this paper simplifies the measurement update process of UKF algorithm and reduces the computation of UKF, based on the establishment of nonlinear state equation and linear measurement equation of CNS/SINS integrated navigation system. Then, we propose one calculation method of attitude covariance matrix in system state covariance matrix based on platform angle error, which derives the expression of platform angle error vector corresponding to prediction error of attitude measurement value in UKF algorithm, and then establishes the UKF algorithm of CNS/SINS integrated navigation system. Finally, the simulation results show that the proposed algorithm can significantly improve the accuracy of navigation parameters of the integrated navigation system as compared with the linear Kalman filter algorithm and EKF algorithm, and that the proposed algorithm has high robustness to the change the initial attitude angle error of the filter.