In order to realize fast, accurate and dynamic prediction of coal and gas outburst in working face, a prediction method based on Principal Component Analysis (PCA) and the weighted Bayesian model is proposed. A weighted Bayesian model for coal and gas outburst prediction in mine working face is built, which also determines the class of the outburst risk. Then the weight of classified variables in the prediction model is determined by using PCA to improve the accuracy of prediction. On this basis, the updating mechanism of the training data is designed based on similarity to rebuild the prediction model effectively. The experimental results show that compared with the naïve Bayesian model and the weighted Bayesian model, the proposed method could quickly generate more accurate prediction results, providing reference for on-site direction of production in mine working face.