The use of computer numerically controlled (CNC) machines has become more widespread and as more machining centers are operating unattended, the need for a Smart CNC machine for on-line tool and process monitoring has become critical. An accurate and reliable method of providing real-time information is vital to the continued integration of adaptive control systems (ACS) with machine tools. ACSs are being developed to monitor parameters like tool wear through current sensing, tool breakage from cutting force signals, and tool chatter from vibration signals. These adaptive control systems' capabilities can be broadened to monitor and control various surface quality parameters. For this to happen, a method to provide accurate on-line information about the machined surface is needed;A multi-level on-line fuzzy net controller and multiple regression model was designed to recognize surface roughness in vertical end-milling process. Both models integrate machining parameters of (1) feed speed, (2) depth of cut, (3) tool type, (4) tool material, (5) work material, (6) spindle speed, (7) vibration, and (8) tool diameter. The fuzzy net controller is composed of eight different fizzy designs each having a fuzzifier, rule base, inference engine, and defuzzifier. Individual designs are referenced to perform surface recognition according to the parameter settings for tool diameter, work material, and tool type;The recognition efficiency of the fuzzy net model and a multiple regression model of same configuration are compared with actual Ra readings taken by a profilometer. This multi-level on-line fuzzy net model displayed a recognition accuracy of 90% as compared to an accuracy of 82% for the multiple regression model.