For the purpose of improving the detecting ability of tool condition change during cutting operation, the time series model parameter obtained from the measured vibration signal and the residual error, which is a difference between the estimated and the measured signals, using genetic programming (GP) are investigated and discussed in relation to wear condition of cutting edge. As a result, it is revealed that the early wear of cutting edge, where the stiffness parameter of the machining part of a cutting model increases linearly with the progress of cutting, can be detected by monitoring of AR parameter of time series model. Moreover, subsequent wear, edge chipping and/or fracture, lead to the increasing of nonlinearity in the measured vibration wave, can be detected by monitoring of residual error. It is therefore concluded that the overall wear behavior in cutting edge can be fully identified by the residual error of the time series vibration signal determined using GP.