The usefulness of a 40-min per trial version of the maintenance of wakefulness test was assessed in 322 patients with obstructive sleep apnea. This test is a variant of the multiple sleep latency test in which patients are asked to remain awake in a quiet darkened room, and then monitored for electroencephalographic sleep onset. The four trials of the test are each stopped after 40 min. The mean sleep latency for all patients was 26.0 +/- 11.8 (SD) min. In a group of 24 patients who underwent treatment with nasal continuous positive airway pressure, the mean sleep latency increased from 18.0 +/- 12.3 to 31.9 +/- 10.4. The strongest nocturnal correlates of the MWT sleep latency were respiratory arousal index (r = -.35), mean oxygen saturation (r = .30), and weight/height ratio (r = -.25). These correlations were comparable to other studies using the MSLT. There were strong intercorrelations among the variables. In the more severe groups, measures of hypoxemia were more strongly correlated with MWT sleep latency. A two-factor analysis of variance using respiratory arousal index and several measures of oxyhemoglobin saturation indicated that both arousals from sleep and degree of hypoxemia contribute interactively to daytime dysfunction in patients with sleep apnea. The MWT appears useful in evaluating disability from daytime sleepiness.