Fourteen ambulatory subjects, varying in their amount of habitual physical activity, were studied for 24h during a total of 25 'typical' days. Rectal temperature was recorded every 6 minutes, an activity diary was filled in every half hour, and wrist activity and heart rate were monitored every minute. Actimetry and heart rate data generally showed close parallelism with each other and with the masking effects on body temperature. Psychological stressors such as public speaking produced a greater effect on heart rate and body temperature than on wrist movement, while typing produced high values for wrist movement, but affected heart rate and temperature much less. When data for the circadian rhythm of body temperature were purified, the diary, actimetry, and measurement of heart rate were all useful in reducing masking effects, but the present evidence indicates that heart rate can be more successful than actimetry - as judged by the closeness of the purified data to a sinusoid. This superiority of heart rate monitoring over wrist activity as a method of purification might be because core temperature can be increased by stressor-induced thermogenesis, as well as by physical activity, and because wrist movement can, with some activities, give an inaccurate estimate of the factors that contribute to whole-body thermogenesis.