A fuzzy-Hammerstein model predictive control method is proposed for a continuous stirred-tank reactor (CSTR). In this paper T-S fuzzy model is used to approximate the static nonlinear characteristics of Hammerstein model, and a linear autoregressive model is used to solve the results of optimal control. The designed nonlinear predictive controller using Hammerstein model make good use of the ability of universal approach nonlinear of T-S model, and divide the question of nonlinear predictive control into the nonlinear model recongnization and the question of linear predictive control. The application results of CSTR process show the proposed control method has good control performance compared to PID controller.