The intrinsic period tau of the circadian pacemaker in humans was investigated by means of forced desynchrony. In this protocol, during 6 scheduled days, the sleep-wake alternation was forced to a period of 20h (i.e., 13.5h for wakefulness and 6.5h for sleep). Light intensity was kept below 10 lux. Three experiments were performed. In experiment 1, 12 subjects were free to spend the available time studying, watching videos, and reading books. In experiment 2, at 2h intervals, 11 subjects spent 6 half-hour sessions per subjective day cycling with minimal effort on a cycle trainer, resulting in an average increase of heart frequency of less than 10 beats per minute. In experiment 3, 11 subjects spent the same intervals of time cycling, but now during 20 minutes per session at an average heart rate of between 140 and 150 beats per minute. Core body temperature was measured continuously. A deconvolution technique discriminated the impact of the circadian pacemaker on body temperature from the impact of activities related to sleeping and waking. From this analysis, the period of the pacemaker was derived. We found the following results: experiment 1, tau = 24.30 +/- 0.36h; experiment 2, tau = 24.17 +/- 0.45h; experiment 3, tau = 23.98 +/- 0.42h. The trend of shorter tau at higher levels of physical activity was not statistically significant. We conclude that tau in humans, determined under conditions of forced desynchrony, is very close to 24h. The suggestion from the literature that single activity pulses would predominantly yield phase delays of the circadian pacemaker is not confirmed by these multiple pulse experiments because no lengthening of tau with increasing effort was observed.