In order to predict the reliability of cutting tools, whose failure is mainly caused by degradation, a forecasting method based on degradation information is proposed, which includes: extraction of degradation indicesycomputation of instantaneous reliabilityyestablishment and application of the neural network prediction model. The key issue of prediction accuracy is the computation of instantaneous reliability. A novel approach incorporated Bayes theorem and Kaplan-Meier (KM) estimator principle is employed to calculate the instantaneous reliability. As validated by the time-varying wear data of the toolsythe trained network is available of predicting the failure time accurately judging by the criterion of reliability. The results show the feasibility and effectiveness to predict reliability based on degradation information.