Abstract: The correspondence of the amount of gas outburst in working face in mine to the geological indexes was researched, and the categories and amount of coal and gas outburst were also analyzed with support vector machine (SVM) method. Then, the SVM model for recognizing two categories of coal and gas outburst, the H-SVMs model for recognizing multi-category of coal and gas outburst, and the SVM regression model for predicting the amount of coal and gas outburst were constructed. The experimental results show that SVM is a good method to recognize the styles of coal-gas outburst;and the SVM regression model is better than the BP model at predicting the amount of coal-gas outburst, because SVM is based on a strict mathematical theory, has a simple structure and a good generalization performance, and can reflect the weights of geological indexes in outburst style recognition through the parameter W in the decision function.