Predicting of port container throughput is characterized by a series of nonlinear factors interaction, which form a high-order, complicated time-varying system. In view of the thought mentioned above, a container throughput predicting method based on SD was proposed. In the model, five subsystems, i.e., environment system, economy system, harbor system, import and export trade system and the collecting & distributing system were included to consider the nonlinear interactions among relative factors. Using this model, the container throughputs from year 2000 to 2020 of Ningbo-Zhoushan port were predicted and the simulated and verified results suggest the model to be effective and feasible.