The Moon and Sleep

Thirty-six per cent of people blame the moon for a bad night. It is the single most-blamed environmental cause, ahead of noise and heat.

They are mostly wrong, and a little bit right, and the shape of the little bit is worth getting exactly.

Three different claims

These get run together constantly. They are not the same claim and they do not have the same evidence.

The first is reasonably solid. The second is shaky and was not replicated. The third is a handful of case reports and the one proper test came back empty.

The claim that holds up

In 2006 thirty-one volunteers in Basel kept sleep diaries for six weeks spanning two full moons. Prospective, designed for the question. They reported 6 h 41 m around the full moon against 7 h 00 m around the new — nineteen minutes, p < 0.001.

Fifteen years later a far better study went into the field. Casiraghi and colleagues put wrist actimeters on ninety-eight Toba/Qom people in Formosa Province, Argentina, in three communities graded by access to electric light, and on 464 students in Seattle. This one was prospective and purpose-built.

Sleep onset moved by twenty to thirty minutes across the cycle. Duration moved by a similar amount. The effect was largest in the communities without electricity — the ones for whom moonlight is still the brightest thing in the sky after sunset — and smaller, but still there, in the city.

That gradient is the strongest single piece of evidence in the whole literature. Turn on the lights and the moon matters less.

The famous study, and what it actually was

The paper everyone cites is Cajochen 2013 in Current Biology: deep-sleep EEG power down thirty per cent around the full moon, melatonin down, time to fall asleep up five minutes, total sleep down twenty.

Read its methods. The authors are admirably direct:

the aim of exploring the influence of different lunar phases on sleep regulation was never a priori hypothesized… We just thought of it after a drink in a local bar one evening at full moon, years after the study was completed.

That is a post-hoc reanalysis of thirty-three people who had come to the lab years earlier for something else entirely, binned afterwards by distance from the nearest full moon. The authors say so plainly, which is more than most papers manage, and it should be quoted rather than politely skipped.

What happened next is the part that never gets repeated.

A group went looking in three large polysomnography datasets — roughly 470, 757 and 870 recordings — and found nothing, publishing it under the title Lunar cycle effects on sleep and the file drawer problem. Then a population cohort in Lausanne put 2,125 adults through full home polysomnography: total sleep time 398 minutes at full moon against 402 and 403 elsewhere, p = 0.31. Their conclusion: "Our large population-based study provides no evidence of a significant effect of lunar phases on human sleep."

So: the timing claim survives. The deep-sleep-architecture claim did not. Anyone quoting "thirty per cent less deep sleep" without the two null replications attached is quoting half a literature.

Moonlight, or gravity?

Casiraghi's own paper argues for moonlight — the electricity gradient points that way, and in the animals where monthly clocks are best understood the cue is unambiguously light. Coral spawn by it. The worm Platynereis can have its monthly clock reset in a tank with nothing but a dim lamp at night.

But their paper also noticed something awkward: a semilunar component, a roughly fifteen-day wobble riding on the thirty-day one. Moonlight has no fifteen-day cycle. Tides do.

In January 2026 the same laboratory posted a reanalysis arguing that the moon's gravitational variation predicts sleep onset better than its illumination — bedtimes about twenty-five minutes later at the gravimetric peak, whichever lunar phase that peak happened to land on — and that in the Seattle cohort there was "no association between the moon's illuminance and sleep onset." They included twenty captive titi monkeys, which have little access to natural light and no idea what a calendar is.

That paper is a preprint. It has not been through peer review. Its authors write that it "is observational in nature and based on statistical correlations, which prevents us from drawing causal inferences," and that the gravitational values "were not directly measured but estimated."

The field is split, and the split is recent, and it is between two analyses by overlapping authors on overlapping data. Nobody should be told the answer yet.

The one to leave out

The lunar-bipolar literature is almost entirely one investigator, largely single patients, entirely retrospective. The one group-level test — nine people, actigraphy and daily mood ratings over one to four months — found "no systematic influence of lunar phase" on sleep onset, sleep duration or mood, and concluded that most people with bipolar disorder show no such pattern.

It is an interesting hypothesis. It is not evidence of anything yet, and it does not belong in a sentence about moon days.

Why people are sure anyway

Cajochen — the same author — went back in 2026 and surveyed 1,815 people about what disturbs their sleep. The moon came top.

Two details give it away. Rumination, far and away the most common actual sleep complaint, was not associated with blaming the moon: the people reporting the classic marker of bad sleep are not the ones pointing at the sky. And moon-blaming peaked in the week after the full moon — after the night in question, when the culprit was available to be remembered.

That is what attribution looks like. A bad night is ordinary and unexplained; the moon is conspicuous and dated; the two get stapled together afterwards.

There is a real effect in here. It is twenty-five minutes of late sleep onset, it is biggest where there is no electric light, and its mechanism is currently being argued about by the people who found it. That is a more interesting thing to know than the folklore, and it is also much smaller.

Sources