Abstract: An efficient method for the estimation of both the state and unknown parameters is presented in this paper. To avoid the results of augmented Kalman filter for both the state and parameters to be divergent, the state and parameter vectors are reconstructured to make them controllable. Based on the controllable models, a U-D covariance factorization filter is developed in which the process noises and observed noises are correlated. This approach has been applied to the estimation of the flight state and instrumental erroros. The results of system simulation and actual application to the aircrafts made in China show that more accurate results can be given by this approach than that given by the augmentted Kalman filter.