A new experimental approach to identify the cutting dynamics in turning operations is presented. This method treats the problem as a closed-loop gray-box identification problem and is based upon spectral analysis and least squares estimation techniques. An active magnetic bearing is employed both to excite the system and to increase the tool’s damping. The inherent time delay of the cutting process is “removed” in identification by limiting the observation period to less than one revolution period of the workpiece. The coefficients of several cutting tests were identified. The methodology was validated by predicting the critical depth of cut from the identification results and comparing with the experiment result measured from a cutting test.