The latency to sleep onset has been reported to be overestimated by chronic insomniacs. Observing that some patients evaluated for suspected hypersomnolence complain of insomnia and others fail to report that they are sleepy, we wondered whether overestimation of sleep latency could be occurring in these subjects as well. Since polysomnography (PSG) only provides one sleep onset with which to assess a patient's estimation, we investigated the use of the multiple sleep latency test (MSLT) for this purpose. Among 147 patients who had an MSLT, 137 of whom had a preceding PSG, overestimation of sleep latency occurred on 78% and 74% of the respective tests. The magnitude of overestimation averaged 3 minutes and 27 minutes, respectively, and was not dependent on diagnosis. Subjects who had reported a history of difficulty falling asleep, compared to those who did not, tended to show equivalent objective sleep latencies, longer subjective nocturnal sleep latencies and less overall accuracy in their estimates. Those who denied having a problem with excessive daytime sleepiness (EDS) showed objective sleep latencies nearly identical to those who complained of EDS but had only a trend toward higher overestimation on the MSLT. Overestimation of sleep latency is therefore more readily part of an explanation for why hypersomnolent patients sometimes complain of insomnia than it is for failure to recognize EDS. The MSLT as well as nocturnal recordings can provide data with which to assess overestimation of sleep latency.