Air pollution is an important and popular topic in Hong Kong as concerns have been raised about the healthimpacts caused by vehicle exhausts in recent years. In Hong Kong, sulphur dioxide SO2, nitrogen dioxide (NO2), nitricoxide (NO), carbon monoxide (CO), and respirable suspended particulates (RSP) are major air pollutants caused by thedominant usage of diesel fuel by goods vehicles and buses. These major pollutants and the related secondary pollutant,e.g., ozone (O3), become and impose harmful impact on human health in Hong Kong area after the northern shifting ofmajor industries to Mainland China. The air pollution index (API), a referential parameter describing air pollution levels,provides information to enhance the public awareness of air pollutions in time series since 1995. In this study, thevarying trends of API and the levels of related air pollutants are analyzed based on the database monitored at a selectedroadside air quality monitoring station, i.e., Causeway Bay, during 1999–2003. Firstly, the original measured pollutantdata and the resultant APIs are analyzed statistically in different time series including daily, monthly, seasonal patterns.It is found that the daily mean APIs in seasonal period can be regarded as stationary time series. Secondly, the autoregressivemoving average (ARMA) method, implemented by Box–Jenkins model, is used to forecast the API timeseries in different seasonal specifications. The performance evaluations of the adopted models are also carried outand discussed according to Bayesian information criteria (BIC) and root mean square error (RMSE). The resultsindicate that the ARMA model can provide reliable, satisfactory predictions for the problem interested and is expectingto be an alternative tool for practical assessment and justification.