A coupled dynamic statistical model (C-TVAREX vector model) is developed for predicting the daily variation of two major pollutants in Macau: the ground-level ozone and the inhalable particulate matter. This model is a Kalmam filter based time-varying autoregressive model with exogenous meteorological inputs, so the model can adapt to the instant change of the meteorological conditions. Different from many ordinary predictive models that can give the forecast of just one type of pollutant, the proposed models in this paper can give forecast for two major pollutants at the same time. Moreover, in order to improve the accuracy of the prediction, the C-TVAREX vector model also considers the interaction between the two major pollutants and the model parameters, i.e., the accuracy of the 8-hr O3 prediction can affect the accuracy of the PM10 prediction, and vice versa. In order to investigate whether considering the interaction between the two major pollutants can enhance the predictive power, the C-TVAREX vector model is compared with another model called Uncoupled TVAREX vector model which is very similar to the C-TVAREX, but does not take the mentioned interaction into account. Results show that the C-TVAREX vector model generally outperforms the Uncoupled TVAREX vector model when forecasting the daily PM10, in terms of generally having higher R, IA, and POD; lower MAPE, RMSE, and PFA. However, the C-TVAREX vector model does not show significant improvement in forecasting the 8-hr O3 in terms of error statistics. Nevertheless, it is suggested that considering the interaction between the predicted pollutants in the TVAREX vector model can improve the predictive power for forecasting the daily PM10 in Macau.