In this study an attempt was made to determine (1) whether the high-amplitude GSR can be suppressed by avoidance conditioning, (2) whether electric shock and red-light warning used as aversive stimuli are equally effective for GSR suppression, (3) whether bio-feedback via TV is more effective than bio-feedback via subject's internal feelings. 　　Thirty freshman students, similar in their GSR wave-patterns, were used as subjects. They were randomly assigned to Shock Group, Warning Group, and Control Group. For the Shock Group, a forearm shock regularly followed a warning signal by two minutes, unless the subjects can intentionally suppress their GSR amplitude to the degree that it is below the base line, within 30 seconds after the termination of the warning signal, and keep it below the line within the remaining 90 seconds. For the Warning Group, the procedure was the same, except that red-light instead of electric shock was used as aversive stimulus. For the Control Group, neither the electric shock nor the red-light warning was used, even if the subjects failed in their avoidance trials. Every subject in the three groups attended the experiment for six consecutive days, receiving 10 avoidance trials each day. In 5 of the 10 trials, the subjects were provided with visual feedbacks of their own GSR record through closed-circuit TV, and in the remaining 5 trials no visual feedbacks were available. 　　The results of this study indicated: (1) that the declining time of the Avoidance Conditioning Group (Shock and Warning combined) was not significantly different from that of the Control Group, (2) that the declining time of the Shock Group was not significantly different from that of the Warning Group, and (3) that the declining time of the TV-feedback Group was not significantly shorter than that of the Self-feedback Group. These results did not seem to support the hypothesis that the autonomic GSRs can be intentionally supressed by avoidance conditioning.