Abstract: A parameter self-adjusting fuzzy weighted controller of multivariable nonlinear system is designed in this paper. First we use fuzzy composed variables to decrease the fuzzy control system dimensions. Then according to the fuzzy composed variable effect on the final decision, we give each fuzzy composed variable a different weight. After self-learning of quantized parameters and weighted parameters with back propagation algorithm, a real-time parameter self-adjusting method based on fuzzy inference system is presented. It gives a solution to the problem for design and on-line realization of high dimension fuzzy adaptive controller. Simulations and real system experiments show its validity.