The study compares the application of the forecasting methods of ARIMA time series and fuzzy time series by Two-factor models, Heuristic models, and Markov models on the amount of Taiwan export. The study compares the forecasting results of different interval length by the three models of fuzzy time series From the conclusion, we find that the methods of fuzzy time series models have better forecasting ability than ARIMA time series model for a short period, namely their mean square error (MSE) is smaller. The Heuristic model is the easiest method to follow. The results suggested that the smaller the interval length is not always the better, and the effective length of interval differed from models.