Abstract: To solve the low prediction accuracy problem of the conventional random forest regression model for the residual life of jig rolling bearings, a method for predicting the residual life of jig rolling bearings combining PCA (principal component analysis) and random forest regression model is proposed. First, the feature set is drawn by time-domain analysis method, and jointly forms a training set with residual life label corresponding to samples; then, PCA algorithm is used to implement dimension reduction processing of the training set features; finally, a random forest regression model is established to predict the residual life of jig rolling bearings. Research results show that: the method based on the combination of PCA algorithm and random forest regression model improves the model prediction accuracy by about 10%, which confirms the effectiveness and accuracy of the method.