Recent reports on the effects of the lunar cycle on seizure occurrence have yielded mixed results. If the moon phase is influential, we hypothesized that this would be due to the moon's contribution to nocturnal illumination, rather than its waxing or waning state, and that significant correlations would not be apparent if local cloud cover were controlled for. We found a significant negative correlation between the mean number of seizures and the fraction of the moon illuminated by the sun (rho=-0.09, P<0.05) in 1571 seizures recorded in a dedicated epilepsy inpatient unit over 341 days. This correlation disappeared when we controlled for the local clarity of the night sky, suggesting that it is the brightness of the night and the contribution the moon phase makes to nocturnal luminance, rather than the moon phase per se, that may influence the occurrence of epileptic seizures.