Four non-consecutive nights of sleep were recorded in eight young volunteers complaining of chronic poor sleep. The subjects received a placebo or a 10 mg zolpidem dose prior to bedtime according to a Latin square double-blind design. All-night spectral analyses of the Cz-Pz lead were associated to the standard polysomnography. According to the visual scoring performed with the Rechtschaffen and Kales criteria, zolpidem significantly increased the stage 4 amount and reduced waking. Compared to placebo no difference in sleep stages was observed when the scoring was based on the power in the 0.7 to 4-Hz band. In zolpidem nights power was significantly reduced in the 4 to 8-Hz band during NREM (stages 2, 3 and 4) and was increased in the 2 to 4-Hz band during REM sleep. A significant reduction of fast activities over 12 Hz was observed during the first 3 h of sleep. Concerning slow wave activity, the only change noted was a significant slowing of its build-up rate at the beginning of cycle 1.