The rapid growth and occasional factors cause fluctuations to time series of number of tourists from mainland China to Macao, especially after the year of 2000, and make it more difficult to model and forecast. The purpose of this thesis is to develop an appropriate Box-Jenkins model to forecast the number of tourists from mainland China to Macao, and attempts to interpret the effects of exceptional external events in order to manage future development when such events happen again in the future. Pre-differencing transformation as well as differencing, together with intervention modeling are applied in this thesis. Box-Jenkins models with pre-differencing transformations are firstly compared. The result shows that models with pre-differencing transformations provide better values of standard error estimate and Ljung-Box statistic than that with original data. In other words, pre-differencing transformation is necessary to stabilize a non-stationary time series with respect to its variance, such as the one analyzed in this thesis, during the process of finding an adequate model. Intervention modeling is then applied to handle exceptional events existing in the time series. Modeling results are compared with those without intervention modeling. The result shows that models with intervention modeling are more adequate in modeling the number of tourists from mainland China to Macao in terms of forecasting accuracy due to obviously improved values not only in standard error estimate and Ljung-Box statistic in general, but also in mean absolute percentage error between forecasting and original data. Application of Box-Jenkins methodology together with intervention modeling is suggested to Macao government for analyzing and forecasting the mainland Chinese tourist arrivals in Macao. Application of intervention modeling technique together with advanced and innovative forecasting methodology is suggested in further study of tourist arrivals in Macao.