This paper is devoted to apply adaptive neural network dynamic surface control (DSC) technique to dual motor driving servo systems.High order neural network is utilized to approximate unknown functions during the design procedure.By defining new state variables,the original system is transformed into state space model with relative order 4.Compared with traditional backstepping method,a first order filter is used to obtain the derivative of the virtual control signal at each step of the controller design procedure in order to eliminate the ”explosion of complexity”.All closed loop signals are proved to be uniformly ultimate bounded by Lyapunov method.Simulation results illustrate the stability of the proposed approach for dual motor driving systems.