This article contains the sleep results of the efficacy study of flurazepam 30 mg and 15 mg, midazolam 15 mg, and placebo in the 99 chronic insomniacs studied as part of this multicenter study. After a 20-day drug washout, all-night sleep was recorded on 2 baseline nights, on the first 2 treatment nights, on treatment night 7, and on the last 2 nights of the study (nights 13 and 14). To reduce the number of comparisons, electroencephalographic (EEG) sleep latency, EEG wake time, EEG sleep efficiency, post-sleep questionnaire (PSQ) sleep latency, and PSQ total sleep were preselected as the major sleep variables. Between-groups comparisons indicated that, when compared with the placebo control, all drugs improved sleep, but differences were statistically significant only for the first 2 nights, i.e., the early interval. Midazolam was more effective than either dose level of flurazepam on treatment night 1. Within-group analyses indicated that all drug groups showed significantly improved sleep from baseline throughout drug administration, but the placebo group did not significantly improve from baseline by either objective or subjective measures at any of the three time intervals. The sleep of patients taking flurazepam 30 mg did not differ significantly from the sleep of those receiving the 15 mg dose for any of the five major sleep variables at any interval. Objective EEG and subjective PSQ sleep variables showed significant positive correlations.