BACKGROUND: Understanding the basis of sleep-related endophenotypes might help to pinpoint factors modulating susceptibility to psychiatric disorders. However, the genetic underpinnings of sleep microarchitecture in humans remain largely unknown. Here we report on the results of a classical twin study in monozygotic (MZ) and dizygotic (DZ) twin pairs examining the genetic effect on sleep electroencephalogram (EEG) composition. METHODS: Polysomnographic recordings were obtained in 35 pairs of MZ (26.4 +/- 5.4 years, 17-43 years, 17 male pairs, 18 female pairs) and 14 same-gender pairs of DZ twins (22.1 +/- 2.7 years, 18-26 years, 7 male pairs, 7 female pairs). The EEG power spectra were generated on the basis of Fast Fourier transformations combined with conventional sleep parameters, according to standardized criteria. RESULTS: We tested the genetic variance contributing to the observed overall variance of the sleep measures and found that the relative contributions of the delta, theta, alpha, and sigma frequency bands at central derivations were significantly correlated to the genetic background. In these frequency bands, MZ twins also showed within-pair concordance in spectral power that was significantly higher than that of DZ twins. CONCLUSIONS: The broad overlap of EEG frequencies during non-REM sleep and wakefulness, which shows a significant genetic variance, supports the hypothesis of common neuronal mechanisms generating EEG oscillations in humans. Our findings strongly support the suitability of the spectral composition of non-REM sleep for defining endophenotypes.