The present study was designed to assess the utility of end-tidal PCO2 (peak concentration of carbon dioxide in a single breath of exhaled air) as an index of psychophysiological activity during performance of a computer-based task and during relaxation. Eleven data-entry operators were monitored continuously for three consecutive, 6 hour work days under the following conditions: (a) during a self-relaxation baseline period; (b) during an abbreviated progressive muscle relaxation period; and (c) during a period of computer-based data-entry work. End-tidal PCO2, respiration frequency, and cardiac inter-beat interval (a measure of heart rate and its variability) were monitored continuously during the three conditions of the study. Self-ratings of relaxation and tension were also monitored at periodic intervals. Consistent with a decrease in psychophysiological arousal, end-tidal PCO2 and self-ratings of relaxation were significantly higher during progressive muscle relaxation than during baseline relaxation. Consistent with an increase in psychophysiological arousal, end-tidal PCO2, cardiac inter-beat interval, and relaxation ratings during data-entry work were significantly lower than during either baseline relaxation or progressive muscle relaxation, while respiration frequency and tension ratings were higher. The findings indicate that end-tidal PCO2 discriminates among different psychophysiological states, and that end-tidal PCO2 may be useful in indexing the stress-health effects of human-computer interactions.