Abstract: The application of adaptive two-stage Kalman filter in transfer alignment is studied. Using the adaptive two-stage Kalman filter, the estimation of the bias is separated from the dynamic state, thereby the size of the matrices and vectors involved in the filtering computations is reduced. The consideration of adaptive filter helps to achieve good performance when the information of the random bias is unknown or incorrect in general. The simulation results show that the misalignment angles of slave navigation system can be estimated fast and accurately by adaptive two-stage Kalman filter.