Tool wear often results in chaotic on milling process. Little research has been carried out on the occurrence and detection of chaotic behavior in time series signal of tool vibration. The vibration acceleration signal based on the operating stages of tool wear was established for the analysis of the correlation dimension of the operating stages of tool wear. Correlation dimension has been calculated to recognize the tool wear operating conditions. Finally, some experimental results from the fractal characteristic showed that there are distinct differences in the correlation dimension in different tool wear conditions and closed the correlation dimension in same tool wear conditions. The correlation dimension can not only be used as an important scientific basis for monitoring tool wear, but also as the complement of other characteristic picking up method.