Previous studies have shown that performance during sleep loss is improved by prophylactic naps as a function of varying nap length. Based on single-dose caffeine studies, a similar dose-response effect has been hypothesized on performance, alertness and mood during sleep loss. The present study compared the effects of repeated versus single-dose administration of caffeine and varying amounts of sleep taken prior to sleep loss on performance, mood and physiological measures during 2 nights and days of sleep loss. A total of 140 normal, young adult males participated at one of two study sites. Ninety-eight subjects at one site were randomly assigned to one of four nap conditions (0, 2, 4 or 8 hours) and 42 subjects at the second site were assigned to one of four caffeine conditions. After a normal baseline night of sleep and morning baseline tests of performance, mood and nap latency, subjects in the nap groups returned to bed at noon, 1600 hours, 1800 hours or not at all. Bedtimes were varied so that all naps ended at 2000 hours. Subjects in the caffeine groups received either a single 400-mg dose of caffeine at 0130 hours each night or repeated doses of 150 or 300 mg every 6 hours starting at 0130 hours on the 1st night of sleep loss. A placebo control group (no nap and placebo administered every 6 hours on the repeated caffeine schedule) was run at both sites.(ABSTRACT TRUNCATED AT 250 WORDS)