The current situation of navigation sensor flight test was briefly analyzed, a method using data fusion algorithm to obtain navigation sensor benchmark was proposed.In this scheme, the performance characteristics of different types of navigation sensors were taken into full consideration.A preliminary test flight navigation data fusion based on centralized Kalman filter was carried out.On the basis of Kalman filtering, fixed-internal smooth filtering was used to achieve global information optimization.The data fusion algorithms were verified by the flight test data.The result shows that, the method makes full advantage of different types of sensors on their characteristics and post data processing capability, and the precision of fusion output is significantly improved.The method supplies a new idea for navigation sensor flight test evaluation.