Due to the influence of uncertain external disturbance and model error, the trajectory tracking of manipulators has some problems such as inaccuracy, low speed and severe chattering. To solve the problems mentioned, a sliding mode control method combined fuzzy controller and Radial Basis Function Neural Network (RBFNN) was proposed. The RBFNN was used to approximate the equivalent system model of manipulators, and the weights of their hidden layers were updated adaptively online. The robust item with the form of sliding mode had been set to compensate the uncertain external disturbances or the approximation errors of the RBFNN. A fuzzy controller was used to adjust the switching gain online adaptively and to solve the chattering of the sliding mode control. The stability and finite time convergence of the closed-loop system had been analyzed by using Lyapunov method. Finally, simulation experiments about trajectory tracking were implemented by MATLAB/SIMULINK, which shows that the proposed method can improve the position-tracking accuracy by more than 40%, the speed-tracking accuracy by more than 70%, and the cumulative-error accuracy by more than 40% compared with the comparison method. Moreover, there is no chattering. The proposed method has been proved to be efficient.