In the process of longterm use, the performance of rolling bearings will degrade to different degrees. If the degradation degree of rolling bearings in work can be quantitatively evaluated, accidents can be avoided.The autoregressive (AR) model was used to filter the vibration signal of the rolling bearing in its whole life cycle.The ratio of residual component energy and signal energy after filtering were calculated,namely AR energy ratio, which was used as a feature vector.The feature vectors were processed by using the standardized min-max normalization method, then they were input to the fuzzy Cmeans (FCM) model. The performance degradation index was obtained and the performance degradation curve was depicted.The envelope spectrum analysis of the signal was carried out, and the correctness of the evaluation results was verified.