Abstract: When LQG/LTR methodology is applied to control system design,it is not easy to choose proper covariance matrices for Kalman filter design for good dynamic performance and robustness of the control system.The methodology is based on the idea that if the control system is decoupled,the target close loop transfer functions of each channel can be presented according to the demands of dynamic performance.After the recovery procedure is finished,the system performance must approximate the target.With the methodology provided for designing the Kalman filter,the Kalman filter gain matrix can be calculated directly in accordance with the performance targets and state space matrices of the plant so that it is unnecessary to solve the Riccati equation.The simulations show that dynamic performance of the control system designed by using this methodology meets the design target,and its robustness is good.