Twelve good sleepers and 12 insomniacs were recorded over three nights in their homes. Subjective assessment of sleep indicated minimal disruption due to sleep recording procedures. A cut point of 88 per cent sleep efficiency resulted in correct classification of nine poor sleepers and 10 good sleepers. Multiple discriminant function analyses, using variables related to rapid eye movement sleep, resulted in the same number of correct classifications. The two indices combined resulted in 100 per cent exclusion of false positives for good sleepers and insomniacs. Insomniacs also showed more night-to night variability in sleep and greater difficulty in returning to sleep once aroused. These results show that good sleepers and insomniacs can be distinquished with a high degree of resolution using variables derived from the home all-night sleep recording.