The Circadian Optimization Protocol: Light + Melatonin + Timing

Key Takeaways

There is a particular kind of tired that no sleep supplement touches. You fall asleep at 1:30 and the alarm goes at 6:45. On Saturday you sleep until eleven and feel human for the first time all week; on Sunday night you lie awake until two and the cycle restarts. Your sleep is not broken. Your clock is set to a different time zone from your life.

This is the problem the Circadian Optimization Protocol was built for. It moves the clock using the inputs the clock actually reads, in the order the evidence supports, and it uses melatonin the way the trials used it — as a timing signal, in a physiological dose, for a few weeks — rather than as a nightly sedative. This article is the reasoning behind the protocol and a four-week plan for running it.

First, the Thing This Protocol Is Not

It is not an insomnia treatment. The American Academy of Sleep Medicine's insomnia guideline suggests against melatonin for chronic insomnia, and the pooled hypnotic effect of the hormone is seven minutes of faster sleep onset. If your body clock is normal and you still cannot sleep, the problem is arousal, not timing, and the answer is cognitive behavioural therapy for insomnia and, on the supplement side, protocols aimed at sleep depth such as the Deep Sleep Stack or Deep Sleep Architect.

How to tell the difference: if you can sleep eight hours soundly whenever you are allowed to choose the time, and the trouble appears only when the world imposes a schedule, your clock is late. If you cannot sleep soundly at any time, it is not.

The Hierarchy, and the Trial That Proves It

Four kinds of signal reach the master clock in the hypothalamus, and they are not equal.

A 2014 randomised trial in 40 young adults with delayed sleep phase makes the practical point better than any table. Participants were assigned to bright morning light or dim light, and to melatonin or placebo — a two-by-two design — and everyone also advanced their rise time gradually. After two weeks the melatonin onset had advanced by about two hours in all four groups. Rise-time advancement alone did it. Light and melatonin only showed their value at three months, when the groups that continued both kept the shift and the others drifted back.

So the protocol has a Layer 0 that is not optional and that costs nothing, and the supplements are placed behind it.

Layer 0: Light and the Wake Time That Does Not Move

Choose a wake time you can hold seven days a week and set it 15 minutes earlier than today's. Every three days, if you are falling asleep earlier, move it another 15 minutes. This is the gradual rise-time advance from the trial above, and it is the engine of the protocol. Weekend lie-ins are the single most common reason the protocol fails: a Saturday that drifts two hours later undoes most of a week.

Get outside within 30 minutes of waking, for 20 to 30 minutes. The international consensus on indoor light sets a daytime target of 250 lux melanopic daylight-equivalent at the eye, and notes that most indoor spaces provide less than 200. An overcast morning outdoors provides thousands. This is the strongest advancing stimulus you have, and its timing — after the core temperature minimum — is exactly where the phase-response curve puts the advance region. If you cannot get outside, a 10,000 lux light box at arm's length for 30 minutes is the clinical substitute.

From three hours before bed, keep light at the eye below 10 lux; in bed, below 1. The median evening light level that halves melatonin is 24.6 lux, and sensitivity varies from 6 to 350 lux between individuals. Overhead lights off, lamps low and warm, screens at minimum brightness with warm filters or off. Amber glasses are a fallback for rooms you do not control, not a substitute — the Cochrane evidence for them is very low certainty, as the light article explains.

Layer 0: Timing Anchors

Close the eating window at least three hours before bed, and keep it consistent. The clock controls glucose tolerance: the identical meal eaten at 20:00 produced 17% higher post-meal glucose than at 08:00 in a controlled laboratory protocol, through a 27% drop in early insulin secretion. A 22:00 dinner compared with 18:00, in 20 healthy adults sleeping the same hours, raised overnight glucose and cortisol and reduced fat oxidation. Early time-restricted eating — a window closing in mid-afternoon — lowered blood pressure by roughly 10 mmHg and improved insulin response in a supervised crossover of eight men with prediabetes, and in 11 overweight adults it reduced mean 24-hour glucose and shifted clock-gene expression within four days.

The counter-evidence tells you which part matters. The TREAT trial randomised 116 adults to a noon-to-8 p.m. window or three structured meals, with no calorie prescription, for 12 weeks: no difference in weight, and a loss of lean mass in the eating-window group. A late, narrow window is not a circadian intervention. An early, consistent one is.

Caffeine stops eight hours before bed. Caffeine equivalent to a double espresso, taken three hours before habitual bedtime, delayed the melatonin rhythm by about 40 minutes in a placebo-controlled study. That is a clock effect, separate from caffeine keeping you awake, and it adds to whatever the evening lights are doing.

Train in the morning or early afternoon while you are advancing. The exercise phase-response curve from 99 adults shows advances with exercise at 07:00 and between 13:00 and 16:00, delays between 19:00 and 22:00. Evening training is not harmful; during the four weeks you are moving the clock, it is working against you.

A warm shower one to two hours before bed. Ten minutes at 40–42 °C in that window shortened sleep onset across a meta-analysis of 13 studies. The body sheds heat afterwards, and falling core temperature is part of how the brain recognises night.

Keep weekends within an hour of weekdays. In 447 midlife day workers, the gap between weekday and weekend sleep midpoints — social jetlag — was associated with lower HDL, higher triglycerides, higher fasting insulin and more body fat, after adjusting for sleep duration and quality.

Layer 1: Melatonin, Used as a Clock Signal

Melatonin enters only if your clock is late, and in two possible timings.

To advance the clock: 0.5 mg in the afternoon, two to four hours before your own melatonin onset. For most people whose sleep comes at 1 or 2 a.m., that is between 5 and 8 p.m. — far earlier than instinct suggests. This is the timing with the largest advances in the 0.5 mg phase-response curve, and at its optimal time 0.5 mg shifted the clock as far as 3 mg did at its own. You will feel mildly drowsy an hour or two later; do not drive after the dose. As your sleep onset moves earlier, move the dose with it.

To fall asleep sooner tonight: 0.5 mg one hour before desired bedtime. This is the regimen from the 116-person delayed sleep-wake phase trial: sleep began 34 minutes earlier than on placebo, and 53% were rated improved versus 24%. In the subset re-measured, the melatonin onset had not shifted more than placebo — the dose was promoting sleep rather than moving the clock. Use this timing if afternoon dosing is impractical, and understand what it is doing.

Why 0.5 mg and not 5. In the dose-ranging study of insomniacs over 50, 0.3 mg restored sleep efficiency and normal night-time blood levels, while 3 mg also helped but caused hypothermia and left melatonin elevated into the next day. Higher doses are stronger sedatives — the meta-regression and the jet-lag review both show it — but for timing the clock, physiological is enough and the side-effect profile is cleaner. The full argument is in the dosing article. Buy a tested product: commercial content ranged from 83% below to 478% above label in a laboratory survey.

Four weeks, then stop. Melatonin here is scaffolding while light and schedule set the new phase. If the phase will not hold without it, Layer 0 is leaking somewhere — usually at the weekend.

Layer 2: If Sleep Depth Is Also the Problem

Many late sleepers also sleep lightly, partly because they are chronically short of sleep. Magnesium L-threonate, 1,500–2,000 mg one to two hours before bed, is the optional layer: magnesium is a cofactor in melatonin synthesis and supports GABA-ergic inhibition. Glycine at 3 g works on the same body-temperature pathway melatonin initiates. If depth is your main problem, move to the Deep Sleep Stack and keep Layer 0 of this protocol running underneath it.

The Four-Week Plan

Week 1 — anchors only. Fix the wake time. Morning light every day. Lights down three hours before bed. Caffeine cut-off. Eating window closed three hours before bed. No melatonin yet. Log sleep onset. For many people, this week alone moves sleep onset 30–60 minutes earlier and the question of melatonin becomes moot.

Week 2 — add melatonin if needed. If sleep onset is still more than an hour later than you need, start 0.5 mg in the afternoon, four to six hours before your current sleep-onset time. Keep advancing the wake time by 15 minutes every three days.

Week 3 — shift the dose with the clock. As sleep onset advances, move the melatonin dose earlier to keep it two to four hours ahead of your (now earlier) melatonin onset. Check the evening light with a phone lux meter at eye level; most people discover a lamp they thought was dim reads 40 lux.

Week 4 — consolidate and withdraw. Hold the schedule. Stop melatonin at the end of the week. If sleep onset holds over the following week on light and schedule alone, you are done. If it drifts, resume for two more weeks and audit the weekend.

The Jet Lag Variant

Eastward across five or more time zones: begin shifting wake time and light exposure an hour a day earlier for two to three days before departure. On arrival, take 0.5–3 mg melatonin at local bedtime for three to four nights, seek bright light in the local morning and avoid it before local dawn. The Cochrane review found a number needed to treat of two, with doses from 0.5 to 5 mg effective and nothing gained above 5 mg. Westward: local evening light does most of the work; melatonin at local bedtime is optional. The AASM practice parameters rate timed melatonin for jet lag as a "standard" recommendation.

Who Should Not Run This Alone

• Shift workers. The principles hold, but the light schedule must be designed around the roster, and the AASM has separate practice parameters for shift work disorder.
• Anyone with suspected sleep apnoea — loud snoring, witnessed pauses, daytime sleepiness on a full night. Fix the airway first.
• Children and adolescents. Delayed sleep phase is common in teenagers and timed melatonin is guideline-endorsed, but paediatric melatonin ingestions reported to US poison centres rose 530% in a decade. A clinician sets the plan.
• People on warfarin, antiepileptics or fluvoxamine. Interactions are documented.
• Pregnant or breastfeeding women. Layer 0 only.
• Anyone with low mood, especially seasonal. Morning bright light roughly doubles remission as an add-on for depression and should be supervised.

Measuring Whether It Worked

The gold-standard phase marker is the dim-light melatonin onset, measured from evening saliva samples in dim light; a sleep clinic can run it, and consensus methods exist for doing so reliably. At home, the proxy is sleep onset time from a diary or wearable, tracked nightly. Expect 15–30 minutes of advance per week when the protocol is working. Track wake-time adherence, minutes of outdoor light before 10 a.m., and a weekly mood score — light moves mood as reliably as it moves sleep.

Frequently Asked Questions

Can I skip the morning light if I use a 10,000 lux box? A light box at the right time is a reasonable substitute. It is still less than an overcast sky, and it does not come with the walk.

What if I take melatonin at bedtime like everyone else? You will get the sleep-promoting effect — about half an hour earlier onset in the best trial — and little phase shift. If your goal is to move the clock, move the dose to the afternoon.

Do blue-light glasses count as "lights down"? Partially. They cut the melanopic component. The Cochrane review could not find convincing sleep evidence for them, and a dark room needs no glasses.

How do I know my own melatonin onset? For most people it falls two to three hours before habitual sleep onset. A clinic can measure it directly. The protocol's timing rules are built on the habitual-sleep proxy, which is good enough for a first pass.

I work from home and my schedule is flexible. Do I need any of this? If you sleep well on your own schedule and your life allows it, no. The protocol exists for the gap between the clock and the calendar.

Where do I start? With the protocol page, which lays out the layers, the doses and the stop rules — and with tomorrow morning's walk.

Related Research

• The Circadian Optimization Protocol
• Melatonin: Optimal Dosing and the Low-Dose Debate
• Circadian Light: Managing Blue Light and Morning Sun
• Melatonin
• The Deep Sleep Stack: Magnesium Threonate + Apigenin + L-Theanine + Glycine