Recently, fuzzy control has been applied to many practical industrial applications. However, as one of the disadvantages of fuzzy control, it is said that we lack, at present, analytical tools for fuzzy control systems. In this paper, a design method which can quantitatively evaluate control performance is proposed for fuzzy optimal control systems. The feature of the design method is to consider not only the dynamics described in consequent parts but also the influence of fuzzy partitions defined in premise parts. The solution of an optimal feedback law consists of two steps. First, a fuzzy system, proposed by Takagi and Sugeno, is approximated by a linear time-varying system. The well-known complex method is used for the approximation since the problem of finding optimum parameters of a linear time-varying system which minimizos the mean squares error criterion is reduced to a non-linear programming. Next, the optimal feedback law is obtained by using the optimal control method for linear time-varying systems. Finally, the effectiveness of this design method is verified by simulation results.