The validity of laboratory-based studies of sleep depends, in part, upon good concordance between habitual sleep schedule and laboratory recording schedule. Without good concordance, error variance due to the circadian misplacement of sleep and to different amounts of time in bed is probable. In an assessment of scheduling concordance in 1,762 research patient nights over two time intervals, we observed good concordance (< 30-minute discrepancy) in 71.2-77.3% of bedtimes and waketimes, discrepancy (difference of > or = 30 minutes) in 14.9-24.2% of bedtimes and waketimes, and missing data in 4.6-7.5% of times. Waketime differences were consistently in the direction of earlier laboratory than habitual waketimes, whereas differences in bedtime were about equally divided between earlier and later (laboratory vs. habitual). Subjects with schedule discordance averaged 19.5 minutes less time in bed during laboratory sessions as compared with their habitual sleep schedule, whereas subjects with schedule concordance averaged only 3.6 minutes less (p < 0.001). Our experience suggests that it may be more difficult to achieve higher rates of concordance among young adult and middle-aged subjects than among elders and that patient requests related to external constraints on scheduling were a frequent reason for discrepancy. We strongly recommend a policy of routinely including data on laboratory versus habitual sleep times in peer-reviewed publications.