To attenuate the effect of uncertainties and unknown disturbances, a composite trajectory tracking control scheme using disturbance observer and neural networks (NN) is proposed for an unmanned surface vehicle (USV) in this paper. In the absence of uncertainties and unknown disturbances, by defining a nonsingular terminal sliding mode (NTSM) manifold, a NTSM-based controller is designed for the USV to guarantee the tracking errors exactly converge to zero within a finite time. In the presence of uncertainties and unknown disturbances, NN is employed to compensate uncertainties while a disturbance observer is applied to simultaneously observe NN approximation error and unknown disturbances. Simulation studies demonstrate the effectiveness of the proposed control scheme.