If machining abnormalities, such as build-up-edge and cutter breakage, occur during machining, serious damage on the workpiece surface and deterioration of production efficiency will happen, and cause additional energy consumption for rework and increase the manufacturing cost. To minimize the effects, an on-line diagnosis method with use of Fast Fourier Transform and algorithm of shorttime signal variation analysis was developed to analyze the vibration signals to quickly detect the two abnormalities in this study. In addition, the real-time machining information from CNC controller was extracted through a bi-lateral communication module to prevent misdiagnosis. The control commands were automatically generated by the proposed system and directly sent to the CNC controller to stop the machine for cutter replacement. Without complex computation, the system can detect the occurrence of BUE or cutter breakage within 1 second and complete machine control within 3 seconds. The system can instantly transmit and save the real machining information and diagnosis/control results to the remote central monitoring platform for further process improvement. With use of TCP/IP communication protocol the central monitoring platform can remote log into the on-site monitoring computer to directly operate the diagnosis/control system. Experimental results showed the feasibility and effectiveness of the proposed system.