Sleep propensity and sleep inertia were assessed in 43 patients with excessive daytime sleepiness (EDS) and 21 sleep-deprived controls, using a forced awakening test under continuous electroencephalographic (EEG) recording. Event-related potentials (ERPs) were first obtained in waking, while participants performed a target detection auditory task. Subjects were then allowed to take a nap with lights off and sleep latency was calculated. After 3 min of continuous sleep, frequent and rare tones were suddenly presented again (and ERPs recorded) in a forced awakening condition, which was repeated a second time if patients fell asleep. ERPs in pre-nap wakefulness did not differ in patients and controls. On forced awakening, almost half (48%) of EDS patients retained morphologically normal ERPs, but showed a significant delay of P300 relative to waking. In the other half of the patients (and none of the controls), the N200/P300 complex to targets was replaced on forced awakening by high-amplitude negative waves ('sleep negativities'). Single subject analysis showed that 65% of patients had abnormal responses during forced awakening (significant P3 delay or sleep negativities), while only three of them (7%) had abnormal ERPs on wakefulness. The presence of sleep negativities was associated with shorter sleep latencies and increased target detection errors on forced awakening. Sleep negativities were more prevalent in narcolepsy and idiopathic hypersomnia than in EDS associated to psychiatric disorders. By combining sleep latency and ERP measures, the forced awakening test provided a robust and relatively rapid tool (45-60 min) to evaluate both sleep propensity and sleep inertia within a single recording session. The test allows each subject to act as his/her own control, thus increasing sensitivity. In the present series, it proved to be much more discriminative than waking ERPs alone to demonstrate specific abnormalities in patients complaining of excessive daytime sleepiness.