The Circadian Optimization Protocol
A protocol that moves your circadian clock with the signals it actually listens to, in order of evidence: a fixed wake time and 30 minutes of outdoor morning light (Layer 0, not optional), an early and consistent eating window with a caffeine cut-off eight hours before bed, dim evenings below 10 lux at the eye, and only then 0.5 mg melatonin — timed in the afternoon to advance a late clock, or an hour before bed to promote sleep onset. Built for people who fall asleep too late, wake unrefreshed, or cross time zones; not a treatment for insomnia, which melatonin does not fix.
Daily Schedule
Timing and dosage for each step
Layer 1 — ONLY if your clock is late: afternoon, 4–6 hours before your habitual sleep onset, weeks 1–4
500 mcg
0.5 mg fast-release, taken 2–4 hours before your dim-light melatonin onset — in practice about 4–6 hours before the time you usually fall asleep, NOT at bedtime. This is the phase-advance timing from the 0.5 mg phase-response curve study (n=34): the largest advances came from afternoon doses, later in the day than the optimum for 3 mg. Expect mild sleepiness within an hour; do not drive. Move the dose 30 minutes earlier every three days as your sleep onset advances. Stop after four weeks and let the light anchors hold the new phase.
Layer 1 alternative — sleep-onset timing: 30–60 minutes before the bedtime you want
500 mcg
The timing used in the 116-person delayed sleep-wake phase trial: 0.5 mg one hour before desired bedtime plus a fixed schedule. Sleep began 34 minutes earlier than on placebo and 53% were rated improved versus 24%. In the subset measured, the melatonin onset did NOT shift more than placebo — this timing promotes sleep more than it moves the clock. Choose this version if you cannot take an afternoon dose safely (driving, machinery, evening work). Adults 55+ with poor sleep maintenance: prolonged-release 2 mg, 1–2 hours before bed, is the form with three-week RCT evidence.
Layer 2 — optional, 1–2 hours before bed, every night
1500 mg
1,500–2,000 mg Magtein (about 144 mg elemental magnesium). Not a circadian agent: magnesium is a cofactor in pineal melatonin synthesis and supports GABA-ergic inhibition during sleep. Include it only if sleep depth, not timing, is also a problem. If depth is your main problem, the Deep Sleep Stack is the better-fitting protocol.
Read This Before the Protocol
The clock is set by light, and almost everything else is a weaker signal. The human phase-response curve to a 6.7-hour pulse of bright light spans about five hours of shift; evening light delays the clock, morning light advances it, and there is no daytime "dead zone" where light does nothing. Melatonin's own phase-response curve runs roughly opposite to light's, with a far smaller amplitude. So the order of this protocol is not a matter of taste: light and schedule first, then meal timing and caffeine, then the tablet.
Melatonin is not an insomnia treatment. The American Academy of Sleep Medicine's 2017 insomnia guideline suggests against melatonin for chronic insomnia; a 19-trial meta-analysis found it shortens sleep onset by 7 minutes. The same academy endorses timed melatonin for delayed sleep-wake phase disorder, and a Cochrane review found a number needed to treat of two for jet lag. If you lie awake at a normal bedtime with a normal body clock, this protocol is not aimed at you. If you fall asleep late, wake late and feel wrong on a conventional schedule, it is.
Two groups should stop here. Children: paediatric melatonin ingestions reported to US poison centres rose 530% in a decade, with two deaths; do not run this on a child without a sleep clinician. Shift workers: the principles apply, but the light schedule has to be built around the shift, and that is a clinician's job.
How This Differs From the Deep Sleep Protocols
This site already has the Deep Sleep Stack and Deep Sleep Architect. They are not competing with this protocol; they answer a different question.
| Deep Sleep Stack / Architect | Circadian Optimization | |
|---|---|---|
| The problem | Sleep is shallow, broken or short | Sleep happens at the wrong time |
| Main lever | GABA, magnesium, cortisol | Light, schedule, melatonin timing |
| Typical person | Falls asleep fine, wakes at 3 a.m. | Cannot fall asleep before 1–2 a.m., dead until noon |
| Compatible? | Yes — low-dose melatonin fits alongside either | Yes — Layer 2 borrows magnesium from them |
If you are not sure which you are, run Layer 0 of this protocol for two weeks first. A clock that is simply late often corrects, and the "insomnia" with it.
The Layer Logic
| Layer | What | Evidence | Optional? |
|---|---|---|---|
| 0 — Light and schedule | Fixed wake time, 30 min outdoor light on waking, evening below 10 lux | A for phase-shifting physiology; B for sleep outcomes | No |
| 0 — Timing anchors | Eating window ending ≥3 h before bed; caffeine cut-off 8 h before bed; exercise before evening | B | No |
| 1 — Melatonin 0.5 mg | Afternoon for phase advance, or 1 h pre-bed for sleep onset | A for delayed sleep phase and jet lag; weak-against for insomnia | Only if the clock is late |
| 2 — Magnesium L-threonate | Sleep-depth support | C | Yes |
The protocol is graded B. The physiology behind Layer 0 is as solid as anything in sleep science, but the outcome trials of light-plus-melatonin in delayed sleep phase are small, and one of them found that a gradual rise-time advance alone produced the same two-hour shift as light and melatonin did in the first two weeks. Honest grade: B, with the weakest link being how much the supplements add on top of the schedule.
Layer 0 — Light and the Fixed Wake Time
Pick a wake time and keep it seven days a week. This is the single intervention that, in a randomised trial of 40 young adults with delayed sleep phase, advanced the melatonin onset by about two hours in every arm — with or without bright light, with or without melatonin. Light and melatonin mattered for keeping the shift at three months, not for making it.
Within 30 minutes of waking, get outside for 20–30 minutes. Outdoor light on an overcast morning is 1,000–10,000 lux; a bright indoor room is usually under 200 lux of melanopic light at the eye. The 2022 international consensus on indoor light recommends at least 250 lux melanopic daylight-equivalent at the eye during the day, and most buildings fail it. In the UK Biobank, each extra hour spent outdoors in daylight was associated with lower odds of depression, easier waking, less tiredness and an earlier chronotype. A week of camping with no electrical light pulled the biological night back to sunset in a classic field study; a single weekend achieved about 69% of that shift.
From three hours before bed, keep light below 10 lux melanopic at the eye; in bed, below 1 lux. Those are the consensus evening and night thresholds. In the dose-response study behind them, 55 young adults exposed to ordinary evening room light showed a median half-maximal melatonin suppression at just 24.6 lux — and individual sensitivity ranged more than 50-fold, from 6 to 350 lux. You cannot know which end you are on without testing, so dim for the sensitive case: warm, low lamps, screens on night mode at minimum brightness, and no overhead lighting after dinner.
Blue-blocking glasses are a maybe, not a must. A 2023 Cochrane review found very low-certainty evidence for sleep quality — three of six small trials positive, three null, no trial measuring melatonin. Amber lenses worn for two hours before bed improved insomnia scores in one 14-person crossover. Wear them if you cannot control the room; do not expect them to substitute for darkness.
Layer 0 — Timing Anchors
Finish eating at least three hours before bed, and keep the window consistent. The same meal eaten at 20:00 instead of 08:00 produced 17% higher post-meal glucose in a controlled laboratory protocol, through a 27% drop in early insulin release. A 22:00 dinner versus 18:00, in 20 healthy adults with identical sleep, raised overnight glucose and cortisol and reduced fat oxidation. Early time-restricted eating (window closing by mid-afternoon) lowered blood pressure by about 10 mmHg and improved insulin response in a tiny supervised crossover of eight men with prediabetes — but a 116-person trial of a noon-to-8 p.m. window with no calorie control changed nothing and cost lean mass. The anchor is early and consistent, not simply narrow.
Caffeine stops eight hours before bed. A double espresso's worth of caffeine taken three hours before habitual bedtime delayed the melatonin rhythm by about 40 minutes in a placebo-controlled laboratory study — roughly half the delay produced by three hours of bright evening light. Caffeine is a clock signal, not just a stimulant.
Exercise in the morning or early afternoon if you are trying to advance. The human exercise phase-response curve, built from 99 adults, shows advances with exercise at 07:00 and between 13:00 and 16:00, and delays between 19:00 and 22:00, with an amplitude comparable to bright light of the same duration. Evening training is not forbidden; it is simply pulling the other way.
A warm shower one to two hours before bed. A meta-analysis of 13 studies found water at 40–42.5 °C for as little as ten minutes shortened sleep onset when timed in that window — the body cools afterwards, which is the signal the brain reads as night.
Layer 1 — Melatonin, Timed to Your Phase
Melatonin earns its place here for one job: moving a late clock earlier. Two timings, two purposes.
To advance the clock: 0.5 mg in the afternoon, two to four hours before your melatonin onset, which for most late sleepers means four to six hours before they actually fall asleep. This is the timing with the largest advances in the 0.5 mg phase-response curve. It will make you a little drowsy mid-evening — that is the point. Shift it earlier as your bedtime moves.
To fall asleep sooner tonight: 0.5–1 mg, 30–60 minutes before the bedtime you want. In the largest delayed-phase trial, this timing plus a fixed schedule brought sleep on 34 minutes earlier than placebo — but did not shift the melatonin onset in those measured. It is a sleep-promoting dose wearing a circadian label.
Why 0.5 mg. In the dose-ranging study of older insomniacs, 0.3 mg restored sleep efficiency and normal night-time blood levels; 3 mg also helped but caused hypothermia and left melatonin elevated into the next day. Higher doses are a stronger hypnotic, and a 2013 meta-regression confirms that — but for timing the clock, physiological is enough. Buy a product you can trust: a lab analysis of 31 commercial brands found content from 83% below to 478% above label.
Run it for four weeks, then stop. Melatonin here is scaffolding. If the new phase does not hold on light and schedule alone, the problem is in Layer 0, not in the dose.
Layer 2 — Magnesium, If Depth Is Also a Problem
Magnesium L-threonate is not a circadian agent. It is here because many late sleepers also sleep lightly, and magnesium is a cofactor in melatonin synthesis and GABA-ergic inhibition. If you need more than this layer, move to the Deep Sleep Stack and keep running Layer 0 of this one.
Jet Lag Variant
Eastward travel across five or more time zones: take 0.5–3 mg melatonin at destination bedtime for the first three to four nights, seek morning light at the destination and avoid light before dawn. Westward: evening light at the destination matters more than melatonin. Doses above 5 mg add nothing; 2 mg slow-release performed poorly in the Cochrane review. Start shifting your schedule by an hour a day for two to three days before departure if you can.
When to Stop and See a Clinician
- You are sleepy during the day on a normal schedule with loud snoring or witnessed pauses. That is sleep apnoea until proven otherwise, not a circadian problem.
- Four weeks of strict Layer 0 and timed melatonin have not moved your sleep onset. A sleep clinic can measure your DLMO and tell you where your clock actually is.
- Low mood, especially winter-pattern. Morning bright-light therapy roughly doubles remission in non-seasonal depression as an add-on; it should be supervised.
- You take warfarin, antiepileptic drugs or fluvoxamine. Case reports and a strong CYP1A2 interaction, respectively, mean melatonin needs a prescriber's sign-off.
- You work rotating or night shifts. The AASM has practice parameters for shift work disorder; the light schedule must be designed around the roster.
Monitoring
| Marker | When | Why |
|---|---|---|
| Sleep onset time (diary or wearable) | Nightly | The outcome. Expect 15–30 min/week of advance if the clock is moving |
| Wake time adherence | Nightly | The anchor. A single late weekend undoes most of a week |
| Minutes of outdoor light before 10:00 | Daily | The dose of Layer 0 |
| Evening light (phone lux meter at eye level) | Weekly spot-check | Target below 10 lux after dinner |
| Daytime sleepiness on a fixed scale | Weekly | Catches too-high or mistimed melatonin |
| Mood on a fixed scale | Weekly | Light moves mood as well as sleep |
Safety and Scope
Melatonin is a supplement in the United States and a prescription medicine in most of the European Union. Short-term use up to three months is where the safety data live; long-term randomised data are scarce. Nothing here treats insomnia disorder, sleep apnoea, depression or shift work disorder. Pregnant and breastfeeding women should run Layer 0 only. Keep melatonin out of children's reach.