The present investigation used a placebo-controlled, double-blind, crossover design to assess the sleep-promoting effect of three melatonin replacement delivery strategies in a group of patients with age-related sleep-maintenance insomnia. Subjects alternated between treatment and "washout" conditions in 2-week trials. The specific treatment strategies for a high physiological dose (0.5 mg) of melatonin were: (1) EARLY: An immediate-release dose taken 30 minutes before bedtime; (2) CONTINUOUS: A controlled-release dose taken 30 minutes before bedtime; (3) LATE: An immediate-release dose taken 4 hours after bedtime. The EARLY and LATE treatments yielded significant and unambiguous reductions in core body temperature. All three melatonin treatments shortened latencies to persistent sleep, demonstrating that high physiological doses of melatonin can promote sleep in this population. Despite this effect on sleep latency, however, melatonin was not effective in sustaining sleep. No treatment improved total sleep time, sleep efficiency, or wake after sleep onset. Likewise, melatonin did not improve subjective self-reports of nighttime sleep and daytime alertness. Correlational analyses comparing sleep in the placebo condition with melatonin production revealed that melatonin levels were not correlated with sleep. Furthermore, low melatonin producers were not preferentially responsive to melatonin replacement. Total sleep time and sleep efficiency were correlated with the timing of the endogenous melatonin rhythm, and particularly with the phase-relationship between habitual bedtime and the phase of the circadian timing system.