The smoothing algorithm for high precise attitude determination in aerial geophysical prospecting is studied to meet with the requirement of high precise attitude and position base. The Kalman filter of 15 states is constructed based on SINS with the aid of single antenna GPS. Analyze the influence of maneuver on state estimation accuracy and compares the effects of Kalman filter with rauch-tung-streibel (R-T-S) smoother through simulation and turntable test. It shows that after R-T-S smooth processing, state estimation accuracy is significantly improved, and high precision references can be provided for aerial geophysical prospecting instrument under long endurance and high dynamic conditions.