This paper presents a robust nonlinear model predictive control (NMPC) scheme for the dynamic positioning (DP) ship subject to time-varying environmental disturbances and input saturation. The control objective is to steer the perturbed DP ship to track prespecified reference trajectories. A nonlinear disturbance observer (NDO) is designed to estimate the unknown disturbances and uncertainties. The NMPC optimization problem is formulated to reject the disturbances by combining the disturbance estimates into the prediction model. The partial asymptotical stability of NMPC is guaranteed without considering any terminal costs and terminal constraints. Furthermore, detail time-iteration NMPC algorithm for DP ships is presented. Simulations of the proposed NDO-based NMPC scheme are implemented for the modes of setpoint regulation and trajectory tracking, respectively. Simulation results well demonstrate the effectiveness and robustness of the proposed control law.