This paper discusses an approach to monitoring cutting tool/part interactions in milling processes using processed acoustic emission (AE) signals and the combined geometric aspects of the tool and the part. The expected events marking interactions of the tool edge with respect to the part during a cut are computed from a priori knowledge of tool and part geometry. Each event period is characterized by its timing and expected AE spectrogram. During cutting, AE signals are periodically converted to spectrograms to determine the peak signal frequency band. This band then defines a bandpass filter through which AE signals pass in real time, reporting the levels and event times of the actual process. Comparison of expected and actual event times and levels forms the basis for a continuous on-line tool/part interaction monitor which detects workholder failure and certain tool failure conditions.