Automatic estimation system of human operator's fatigue is considered valuable for prevention of industrial accidents and hazards as well as improvement of operational efficiency, if it can be realized. In the system, from the standpoint of practical use, the information required for the estimation must be the operational and physiological ones, which can be detected easily with little disturbance of operator's action. In this research, the objective task is small hole drilling and the authors have chosen fluctuation of operator's body motion, cutting force, task time, task error, blood pulse period, respiration period, electromyogram, electroencephalogram, etc. as the information. And then, the relation between the information and fatigue has been investigated by using the multiple regression analysis. The fatigue has been evaluated by the critical flicker fusion value and subjective fatigue symptom point. As the result, it has been found that there was a close relation between the detected information and fatigue for every operator individually, though a unique regressive equation for all operators could not be found.