Abstract: This paper describes a method for extracting the aerodynamic derivatives of a helicopter from flight data by means of low pass filtering, Kalman filtering and least square techniques. The method can be summaried as follows:a ) The high frequency effects from the rotor are eliminated by low pass filtering, and then measurement noise and process noise statistics are obtained.b) By using Kalman filter, the random noise is minimized and the bias error is eliminated, but the derivatives are not identified yet.c) The derivatives are identified from raw data by least square technique. They serve as initial values for Kalman filtering.d) The final derivatives are extracted from Kalman filtering data by least square technique.The method presented requires considerably less computation than the maximum likelihood method of reference [ 1 ]. It is more accurate than the least square technique and the least square technique with Kalman filtering data of reference [ 1 ]. This method cut down more than 70 percent of error in comparison with the least square technique.