Abstract: The measure of the predictability with historical time-series data of port throughput is one of uncertain issues caused by the strong randomness of the data. It is suggested that the entropy might be a measure for the predictability. This concept is checked out through examining the container throughput time-series data from 20 ports in China. SampEn（Sample Entropy） is used to measure the complexity of the time series data and the ARIMA（Autoregressive Integrated Moving Average）model is used to calculated the container throughput prediction of the port. The research shows that the correlation between SampEn and the predictability is rather weak and the ARIMA model is only good for large ports or growing ports.