Low reliability and cumbersome calibration procedures for commercially available drill breakage detection system were the drivers for the development of a robust system which utilizes time and frequency domain analysis of vibration signatures from the spindle housing. Self learning capabilities in calibration and generic, multidiscriminant based decision making are the novel features of a system proven successful in single spindle applications1. However, use of a single sensor to monitor drill breakage in multi spindle station in a high volume manufacturing operation requires signal enhancement strategies to decipher similar signatures sensed from different spindles. Complexity of the problem increases if the station is one of the several stations in a dial machine, because one needs to consider the transmissivity characteristics between stations installed on a common rotary table. Vibration signatures generated due to impacts associated with part clamping/unclamping, slides positioning and retracting, drills piercing in through holes, etc. reflect non-coherent components which have similar characteristics as associated with drill breakage. Accordingly, single domain analysis in the time or the frequency domain is insufficient to detect tool failure. An application of discriminant enhancement strategies and algorithms validation is described for the use of a single sensor to monitor tool breakage in a dial machine station with two spindles drilling parts at the same speed. Stability of the discriminants is addressed through statistical analysis, which can be used not only for detecting abnormalities in the machining process, but also for developing algorithms to predict tool wear in drilling operation.