Although there is good consensus that a single administration of melatonin in the early evening can phase advance human circadian rhythms, the evidence for phase delay shifts to a single melatonin stimulus given in the early morning is sparse. We therefore carried out a double-blind randomized-order placebo-controlled study under modified constant routine (CR) conditions (58 hr bedrest under approximately 8 lux with sleep 23:00-07:00 hr) in nine healthy young men. A single (pharmacological) dose of melatonin (5 mg p.o.) or a placebo was administered at 07:00 hr on the first morning. Core body temperature (CBT) and heart rate (HR) were continuously recorded, and saliva was collected half-hourly for assay of melatonin. Neither the timing of the mid-range crossing times of temperature (MRCT) and HR rhythms, nor dim light melatonin onset (DLMOn) or offset (DLMOff) were phase shifted the day after melatonin administration compared with placebo. The only change was an altered wave form of the CBT rhythm: longer duration of higher-than-average temperature after melatonin administration. Under the same modified CR conditions we have previously demonstrated a clear phase advance of the above circadian rhythms following a single administration of 5 mg melatonin in the evening. This study's failure to find significant delays to a single administration does not negate other positive findings with multiple doses, which may be necessary for a 'weak zeitgeber'.