To address the issue of reduced filtering accuracy in integrated navigation systems caused by variable noise interference, an algorithm based on interactive multi-model (IMM) and square-root cubature Kalman filter(SCKF) is proposed. The IMM-SCKF filtering algorithm employs multiple model sets and adjusts the probability of the sub-model while fusing the output, allowing it to simulate the actual noise covariance to a certain degree. Simulation and road test results show that the root mean square (RMS) error of the IMM-SCKF algorithm is superior to that of the traditional single-model CKF algorithm, effectively enhancing the reliability of the integrated navigation system. Compared to the traditional CKF algorithm, the IMM-SCKF algorithm reduced the RMS error in eastward, northward, and up speed errors by 52%, 55%, and 30%, respectively, and the RMS error in position by 47%,60%, and 32%, respectively. The IMM-SCKF algorithm significantly improves the positioning accuracy and antiinterference ability of the system.