This paper describes the development of a sensor incorporating an algorithm that estimates the quality of comfortableness by measuring peripheral skin temperature, pulse, and galvanic skin response (GSR) that reflect autonomic nervous system activity. A correct answer rate of 83% was obtained between the subjective comfortableness and the estimated comfortableness by the developed sensor. For the application of this human feeling sensor, we used it to estimate the comfortableness of subjects receiving massages, then developed two alternative adaptive massage control procedures based on the sensor's estimate, and verified the effectiveness of the results through testing the subjects. We observed that in course A (dynamic comfortableness) the peripheral skin temperature dropped and the GSR and pulse rate increased, while in course B (static comfortableness) the peripheral skin temperature rose and the GSR and pulse rate dropped. By the end of the control sequence, there was a statistically significant difference in the amounts of change in both the peripheral skin temperature and the GSR between the two sequences (p < 0.05). To examine the effects of each course more closely, we mapped the trajectories of physiological change during the control sequences of each course at twentysecond intervals, and the results correlated closely with the subjective assessments. These results suggest that bio-control adapted to comfortableness is feasible.