Tool monitoring is an important factor to restrict the improvement of production efficiency, machining quality and automation level. The monitoring of the tool wear and breakage conditions on YCM-V116B machining center was studied, and the acquired milling sound signals were analyzed in detail. By means of the classical time-frequency analysis, it was discovered that the wear sound had its own characteristic frequency band, and the frequency component within the frequency band would change according to the change of wear conditions. So that the frequency component within the frequency band will be a good indicator to monitor the tool wear conditions excellently. On the other hand, the tool breakage sound is a random signal that a transient change in amplitude is produced probably when tool breaks. The tool breakage conditions can be detected exactly by the advantages of wavelet decomposition techniques. The analysis implies that the sound generated during the machining process can be used to monitor tool conditions, which provides a new approach to the sound applications in tool monitoring domain.