Endogenous rhythms of circalunar periodicity (similar to 29.5 days) and their underlying molecular and genetic basis have been demonstrated in a number of marine species [1, 2]. In contrast, there is a great deal of folklore but no consistent association of moon cycles with human physiology and behavior [3]. Here we show that subjective and objective measures of sleep vary according to lunar phase and thus may reflect circalunar rhythmicity in humans. To exclude confounders such as increased light at night or the potential bias in perception regarding a lunar influence on sleep, we retrospectively analyzed sleep structure, electroencephalographic activity during non-rapid-eye-movement (NREM) sleep, and secretion of the hormones melatonin and cortisol found under stringently controlled laboratory conditions in a cross-sectional setting. At no point during and after the study were volunteers or investigators aware of the a posteriori analysis relative to lunar phase. We found that around full moon, electroencephalogram (EEG) delta activity during NREM sleep, an indicator of deep sleep, decreased by 30%, time to fall asleep increased by 5 min, and EEG-assessed total sleep duration was reduced by 20 min. These changes were associated with a decrease in subjective sleep quality and diminished endogenous melatonin levels. This is the first reliable evidence that a lunar rhythm can modulate sleep structure in humans when measured under the highly controlled conditions of a circadian laboratory study protocol without time cues.