The aim of the present study was to evaluate the sensitivity of the human circadian pacemaker to the resetting effect of moderately bright light (approximately 1260 lux), and to assess the direct effect of such light exposure by comparison to a control group of subjects undergoing the same behavioral manipulations but with a similarly timed exposure to darkness instead of light. Endogenous circadian phase and amplitude were assessed in dim light (approximately 10-15 lux) before and after two consecutive series of three 5-hr exposures to approximately 1260 lux or to darkness (approximately 0.03 lux) in two different groups of young healthy men, using the constant-routine technique. The light or darkness exposure was centered 1.5 hr after the initial fitted endogenous temperature minimum and 12 hr opposite the newly scheduled midpoint of the sleep episode, in order to induce a phase advance in the light-exposed subjects. The phase of the endogenous circadian pacemaker was assessed by a dual-harmonic regression model from core body temperature recorded every minute during constant routines. Urinary volume was measured at each micturition, subjective alertness every 20 min, and cognitive performance hourly. The endogenous circadian phase shifted to a significantly earlier time after each series of light exposures in the treatment group than it did in the control group (two-way analysis of variance for repeated measures: F = 67.91, p = 0.0001). The analysis of circadian curves of urine production, subjective alertness, and cognitive performance scores revealed that all variables maintained stable temporal relationships with the endogenous circadian temperature minimum--an indication that these rhythms shifted in the same direction and by an equivalent amount. Despite comparable behavioral schedules, including the timing of bedrest/sleep and social contacts, circadian temperature rhythm of control subjects free-ran under dim light conditions, indicating that moderately bright light exerted a direct biological effect on the human circadian pacemaker in the treatment group. The present study also demonstrated that light of approximately 1260 lux (which is of substantially lower intensity that the approximately 7000-12,000 lux used in prior experiments) produces robust phase advances of the endogenous circadian temperature rhythm, which has been shown to be an accurate marker of the output of the circadian pacemaker (Czeisler et al., 1989). These results support the hypothesis that the phase-shifting effect of light on the human circadian pacemaker has a strongly nonlinear relationship to illuminance levels, such that it is preserved despite marked reductions in light intensity.