Suspended particulate matters (PM10) is considered as a harmful air pollutant. Many models attempt to predict numerical levels of PM10 but a simple, clearly defined classification of PM10 levels is more readily comprehensible to the general public rather than a numerical value. However, the PM10 prediction model often suffers from data imbalance problem in the training dataset that results in failure to forecast the minority class of severe cases. In this study, a warning system using extreme learning machine (ELM), compared with support vector machine (SVM), was constructed to forecast the class of PM10 level: Good, Moderate, and Severe. An imbalance strategy called prior duplication was also applied to improve the forecast of minority class. The experimental comparisons between ELM and SVM demonstrate that ELM produces superior accuracy relative to SVM in forecasting minority class (Severe) of PM10 level with or without the imbalance strategy. Furthermore, our results show that the required training time and model size in the ELM model are much shorter and smaller than those of SVM respectively, leading to a more efficient and practical implementation of prediction model for large dataset. The performance superiority of ELM is also discussed in this paper. © 2013 Elsevier B.V.