Abstract: In view of problem that inner mechanism of coal and gas outburst was very complex and it was difficult to establish an appropriate multivariate nonlinear prediction model, prediction method of coal and gas outburst based on catastrophe progression method was proposed. The method chooses coal and gas outburst instances of typical outburst mine as learning samples to determine catastrophe progression with different levels of impact risk, and uses them to predict the other samples. The actual application results show that the method can accurately reflect risk degree of coal and gas outburst and has high prediction accuracy.