Polysomnograms were obtained from 37 volunteers, before (baseline) and after (two consecutive recovery nights) a 64-h sleep deprivation, with (d-amphetamine or modafinil) or without (placebo) alerting substances. The drugs were administered at 23.00 hours during the first sleep deprivation night (after 17.5 h of wakefulness), to determine whether decrements in cognitive performance would be prevented; at 05.30 hours during the second night of sleep deprivation (after 47.5 h of wakefulness), to see whether performance would be restored; and at 15.30 hours during the third day of continuous work, to study effects on recovery sleep. The second recovery night served to verify whether drug-induced sleep disturbances on the first recovery night would carry over to a second night of sleep. Recovery sleep for the placebo group was as expected: the debt in slow-wave sleep (SWS) and REM sleep was paid back during the first recovery night, the rebound in SWS occurring mainly during the first half of the night, and that of REM sleep being distributed evenly across REM sleep episodes. Recovery sleep for the amphetamine group was also consistent with previously published work: increased sleep latency and intrasleep wakefulness, decreased total sleep time and sleep efficiency, alterations in stage shifts, Stage 1, Stage 2 and SWS, and decreased REM sleep with a longer REM sleep latency. For this group, REM sleep rebound was observed only during the second recovery night. Results for the modafinil group exhibited decreased time in bed and sleep period time, suggesting a reduced requirement for recovery sleep than for the other two groups. This group showed fewer disturbances during the first recovery night than the amphetamine group. In particular, there was no REM sleep deficit, with longer REM sleep episodes and a shorter REM latency, and the REM sleep rebound was limited to the first REM sleep episode. The difference with the amphetamine group was also marked by less NREM sleep and Stage 2 and more SWS episodes. No REM sleep rebound occurred during the second recovery night, which barely differed from placebo. Hence, modafinil allowed for sleep to occur, displayed sleep patterns close to that of the placebo group, and decreased the need for a long recovery sleep usually taken to compensate for the lost sleep due to total sleep deprivation.