In order to realize the intelligent machine tools, which can autonomously determine the cutting states and can change them automatically as required due to changes in the environmental conditions, an in-process monitoring and identification of cutting states is developed for CNC turning machine to check and improve the stability of the processes. The method developed utilizes the power spectrum density, or PSD of dynamic cutting force measured during cutting. Experimental results suggested that there are basically three types of patterns of PSD when the cutting states are the continuous chip formation, the broken chip formation, and the chatter. The broken chip formation is desired in order to realize safe and reliable machining. The proposed method introduces three ratios, which are calculated from three dynamic cutting force components and obtained by taking the ratio of cumulative power spectrum density for a certain frequency range corresponding to the states of cutting to that of the whole frequency range of each dynamic cutting force component, to classify the cutting states of continuous chip formation, broken chip formation, and chatter. The algorithm was developed to calculate the values of three ratios during the process in order to obtain the proper threshold values for classification of the cutting states. The method developed has been proved by series of cutting experiments that the states of cutting are well identified regardless of the cutting conditions. The broken chips are easily obtained by changing the cutting conditions during the processes referring to the algorithm developed.