PURPOSE: To examine whether nocturnal exercise with body temperature (Tb) increase would accelerate circadian phase delay compare to exercise with blunted increase of Tb. METHODS: Seven healthy collegiate men who had not smoked and had no mental or endocrine diseases (20.5±2.9 yrs, 174.4±4.0 cm, 70.1±6.1 kg, 10.7±1.9% body fat) were recruited. Subjects participated in two separate (at least 15 days), counterbalanced, experimental sessions, and each session lasted for 5 days. On day 1 of each session, subjects maintained habitual sleep time (2300–0700, 0.2 lux) in the laboratory. From day 2 thru 5, they exercised on a stationary bike for 60 min at 55% of maximal heart rate (HR) beginning at 0100 (15 lux), and subsequently bedtime and waketime were fixed to 0400 and 1200, respectively. In one session, they cycled at an ambient temperature of 26°C with Tb increase (HT, average Tb 36°C) while they did at 17°C with cooling devices for blunted Tb increase (LT, average Tb 33.5°C). During exercise, rectal (Tre) and skin (Tsk) temperatures, and HR were continuously monitored and recorded. Body weight was measured before and after each exercise. Urine volume was measured as they urinated. Blood samples were collected at 2300, 0330, 0700, and 1200 hours on day 1 and 5 of each session and analyzed for melatonin. RESULTS: The average weight loss for 4 days of nocturnal exercise in HT and LT was 0.62±0.09 and 0.22±0.07 kg, respectively (P<0.001). Tre increased approximately 0.6°C at the end of each exercise bout for both HT and LT. Tsk maintained at 32°C and 24°C during exercise for HT and LT, respectively (P<0.001). The average Tb was higher in HT than LT during exercise. HR was significantly elevated during exercise in HT (average 120 bpm) than LT (average 112 bpm) (P<0.05). The average total urine volume passed between 0100 and 1200 for 4 days was 0.07±0.07 in HT and 0.11±0.07 liter inLT (P<0.05). The melatonin concentration at day 1 was 23±26, 107±45, 98±46, and 14±5 inHT and 18±10, 108±65, 103±75, and 14±12 pg/ml in LT for each time period. At day 5, it was 9±3, 64±41, 122±73, and 54.1±17.8 inHT and 8±1, 68±21, 111±52, and 32±14 pg/ml in LT for each time. The melatonin level at 1200 was higher in HT than LT (P<0.05). CONCLUSIONS: The results suggest that varied whole body thermal stress during nighttime may contribute to modify circadian phase shift. In the present study, increase of Tb during nocturnal exercise produced a marginal phase delay compared to blunted Tb increase during exercise.