Sleep scoring of whole-night polysomnograms is labor intensive. Scoring fewer epochs saves labor at the cost of accuracy; this study investigates the trade-off between the two. Whole-night sleep measures of 12 patients with sleep apnea syndrome, 10 patients with narcolepsy, and 35 controls were first computed using conventional successive 30-second epochs. Using the resulting list of sleep stages, a variable number of epochs was skipped among remaining epochs; the measures were recomputed for the reduced lists. The Bland-Altman analysis was used to define the agreements among the sleep measures at the conventional resolution and those at the lower resolutions. Scoring one-half to one-third of the number of epochs changes the duration of sleep stages only up to 2.5% and 5%, respectively, for all groups and sleep stages. In apnea patients, rapid eye movement (REM) latency deviates < 15 minutes when half of the epochs are scored. In controls and narcoleptics, much lower resolutions can be used before reaching the same level. Potential restrictions for the application of the method are discussed.