Since the trajectory sensitivity can not be modeled accurately in a closed-loop formulation, the linear preregulator design techniques are investigated to approximate the sensitivity information, and two methods leading to transfer-function matrix form are developed. The obtained control systems have shown much improvement in reducing the sensitivity, specially for Method B. The advantages of Method B over Method A are that the difficulty of obtaining the steady state control law in terms of the reference input can be prevented and that the controller containing an integral action can be obtained to ensure that there is no steady-state error.