Abstract: The paper proposed a forecasting method of coal and gas outburst based on KPCA-SVM. The method firstly used KPCA to select features of correlative indexes influencing coal and gas outburst, then used SVM to make classified forecasting coal and gas outburst. The example forecasting result showed that the correctness rate of forecasting coal and gas outburst by the method is obviously higher than the one by SVM, and the method has quick operating speed and strong identification ability. Meanwhile, the forecasting model of coal and gas outburst on the basis of the method has good stability and validity.