The wear volume is obtained by means of experiment and the wear random process model is established according to the result. The Carbon steel material is used and the samples are grouped five after surface treatment, each group tests six times under the same condition. The wear volume under each wear time shows big dispersion. The additional study indicates that the sample has the large wear volume is in the serious wear state from the beginning, and the wear of running-in phase is inflected by the work velocity and the state condition of the surface of the samples. The wear process which the mean value is a constant and the standard deviation is different is a normal process generally, it is a stationary normal process if the standard deviation has no relation to the start of the wear time, or a Wiener process if the standard deviation is liner with the wear time, it is valuable to forecast the wear reliability.