This paper discusses the monitoring of cutting tool wear based on time series analysis of acoustic emission signals. In cutting operations, acoustic emission provides useful information concerning the tool wear condition because of the fundamental differences between its source mechanisms in the rubbing friction on the wear land and the dislocation action in the shear zones. In this study, a signal processing scheme is developed which uses an autoregressive time-series to model the acoustic emission generated during cutting. The modeling scheme is implemented with a stochastic gradient algorithm to update the model parameters adoptively and is thus a suitable candidate for in-process sensing applications. This technique encodes the acoustic emission signal features into a time varying model parameter vector. Experiments indicate that the parameter vector ignores the change of cutting parameters, but shows a strong sensitivity to the progress of cutting tool wear. This result suggests that tool wear detection can be achieved by monitoring the evolution of the model parameter vector during machining processes.