The forecasting of container volumes is a critical component in both the nation's economy and the development of ports. This paper attempts to generate a model for forecasting container volumes at ports using system dynamics to provide higher forecasting accuracy. First, a regression model and system dynamics model are employed to forecast the container volume in Taiwan, the average forecasting error shows that the system dynamics model has higher forecasting ability. Second, data for the period 1990-2008 in Taiwan and Korea are used to forecast the container volumes for 2016. The total container volumes come from the import/export container volume and transshipment container volume. The input data are divided into two sets-total and import/export-according to the properties of the port or nation. By comparing the error of the result using the two sets of input data, the system dynamics model in this paper is improved and is shown to be a good method for forecasting container volumes at different ports or nations in the world.