Quick answer: A randomised trial moved self-identified night owls about two hours earlier in three weeks, using four everyday levers and no drugs: a fixed wake time with morning light, fixed meal times, no caffeine after early afternoon, and exercise moved earlier in the day. Sleep duration did not shrink, and the participants’ mood and morning reaction times improved. That is the realistic shape of this: roughly 30 to 60 minutes per week, not one heroic night. The single highest-leverage action is a fixed wake time you keep at weekends — in 60,977 people wearing accelerometers, how regular sleep was predicted death more strongly than how long it was.
It is 2am. You have work at eight, you have been lying here since midnight, and you have already done the arithmetic on how many hours you can still get. You are not looking for a lecture about sleep hygiene. You want to know how to turn this around and how long it is going to take.
Here is the honest version, with the numbers attached.
Regularity matters more than you have been told
Nearly every piece of sleep advice is about duration — the eight hours. The evidence has moved.
A 2024 study in Sleep analysed more than 10 million hours of accelerometer data from 60,977 UK Biobank participants, scoring each person on a Sleep Regularity Index — how consistent their sleep and wake timing was day to day. Over an average 6.3 years of follow-up, the four most regular quintiles had a 20% to 48% lower risk of all-cause mortality than the least regular quintile, with similar patterns for cancer (16–39% lower) and cardiometabolic death (22–57% lower), adjusted for age, sex, ethnicity, lifestyle and health factors.
The finding that matters for you: sleep regularity predicted mortality more strongly than sleep duration did.
A separate systematic review covering 41 studies and 92,340 participants across 14 countries reached the compatible conclusion: later sleep timing and greater variability in sleep were both associated with worse health outcomes, and consistent bedtimes and wake-up times with better ones.
This is good news if your schedule is a mess, because consistency is a far easier target than a magic number of hours. It also reframes the goal. You are not trying to sleep more tonight. You are trying to make your nights look the same as each other.
If you are reading this at 2am because you cannot get to sleep, the one thing with real circadian evidence is low-dose melatonin taken in the evening — not the 5 or 10mg gummies, which are far above the dose used in phase-shifting research. YU SLEEP is dosed at 0.9mg, which sits in the right range, though the minimum order is $138 and that is a real consideration — our full review covers who should skip it. If you have never tried plain low-dose melatonin on its own, do that first; it costs a few dollars and tells you something useful.
What actually moved the clock in a trial
Most schedule advice is a list of twenty things. A 2019 randomised controlled trial in Sleep Medicine is more useful because it tested a specific, practical bundle in the real world — not a sleep lab — on people whose body clocks ran unusually late.
Participants were asked to do four things:
- Wake up 2–3 hours earlier than usual and get outdoors light, then go to bed 2–3 hours earlier, keeping it consistent on weekdays and weekends
- Eat breakfast as soon as possible after waking, lunch at the same time each day, and nothing after 7pm
- No caffeine after 3pm
- Exercise only in the morning, on a fixed routine
Measured by actigraphy and by two independent circadian markers — dim light melatonin onset and the peak of the cortisol awakening response — the group advanced their sleep and wake timing by about two hours, with no reduction in sleep duration. Alongside it came significant improvements in self-reported depression and stress, and better reaction time and grip strength during the morning hours that had previously been their worst.
Two things to take from that. The bundle works, and it works in about three weeks. Nothing in it is exotic, and nothing in it is a supplement.
The 14-day version
Compressed into something you can start tomorrow. The order matters: the morning anchor comes first, because everything else hangs off it.
Days 1–3: fix the wake time only
Pick a wake time you can hold seven days a week and set it. Do not touch your bedtime yet — you will not fall asleep earlier by deciding to, and lying in the dark failing is its own problem. Get outside within 30–60 minutes of waking, for 15–20 minutes if you can. Outdoor light on an overcast morning is still many times brighter than indoor lighting.
Expect these three days to be unpleasant. You are deliberately accumulating a little sleep pressure, and that is the mechanism, not a side effect.
Days 3–7: add the evening
Now dim the lights in the last 90 minutes and get the screens down. The size of this effect is not trivial — see the next section. Move caffeine to before early afternoon. Eat your last meal earlier, and keep lunch at a consistent time, since meal timing is one of the levers the trial used.
Your bedtime should start drifting earlier by itself around here. Let it. Do not force it.
Days 7–14: hold it, including the weekend
This is where most attempts die. Sleeping in on Saturday by two hours is the equivalent of flying two time zones west and back every week, which is why researchers call it social jetlag, and the systematic review above found it consistently associated with worse outcomes.
If you are short on sleep by Saturday, a short early-afternoon nap costs you far less clock than a lie-in does. Keep the wake time.
After day 14
On the trial’s timeline you are about two-thirds of the way to a two-hour shift. Keep going. If you have moved nothing at all in two weeks of genuinely holding the wake time, that is worth a conversation with your doctor rather than more willpower — see the last section.
Screens: the actual measured effect
Everyone knows screens are bad before bed. Fewer people know by how much, and the number is more motivating than the advice.
Researchers at Brigham and Women’s Hospital and Harvard Medical School compared reading a light-emitting e-reader before bed against reading a printed book, in a controlled study published in PNAS. Readers on the light-emitting device took longer to fall asleep, felt less sleepy in the evening, secreted less melatonin, had their circadian clock shifted later, and were less alert the next morning.
That is the direction you are fighting when you scroll at 11pm: the device is actively pushing the clock the way you are trying not to go. Dimming the room and putting the phone down for the last hour is not folk wisdom — it is reversing a measured effect.
Where melatonin actually fits
Low-dose melatonin in the evening is the one supplement with genuine phase-shifting evidence behind it, and it is widely misused. Two mistakes dominate.
The dose is usually far too high. Supermarket melatonin commonly comes at 5mg or 10mg. The doses used in circadian research are typically a fraction of that. More is not better here, and higher doses are associated with more next-day grogginess without more phase shift.
The timing is usually wrong. Melatonin taken to shift your clock earlier is an evening signal taken some hours before your target bedtime, not a sedative you take as you get into bed. Taken at the wrong time it can push your clock the wrong way.
Morning light does the matching job from the other end. In a small study of adults with type 2 diabetes, 30 minutes of 10,000 lux light each morning for two weeks advanced melatonin onset by around 23 minutes and improved daytime sleepiness — a modest, measured shift from a cheap intervention, though with only ten participants it is a signal rather than proof.
For the nights while the reset catches up
The reason we point people at YU SLEEP for this specific job is the dose: melatonin at 0.9mg, in the range the research uses, in a sublingual liquid rather than a 10mg gummy. It is for the rough nights while your clock catches up to your new wake time — not a nightly habit, and not a substitute for the morning light that does the actual work.
From $138 for two months · no single-bottle option · 60-day money-back guarantee · full review
Check first: YU SLEEP contains 5-HTP, which raises serotonin. If you take an antidepressant (SSRI, SNRI or MAOI), a migraine triptan or tramadol, ask your prescriber before trying it. Not for pregnancy or breastfeeding.
What does not work
Staying up all night to “reset”. The single most common piece of internet advice on this, and the one most likely to leave you worse off. It is covered properly in can you fix your sleep schedule in one night.
Sleeping in to catch up. It feels like repayment and it is the thing most reliably holding your clock where it is.
Going to bed earlier as the first move. Your clock is not ready, so you lie there, and your brain starts associating the bed with being awake and frustrated. Fix the wake end first.
High-dose melatonin every night. Covered above.
When it is not just a messy schedule
Some of this is not a discipline problem, and treating it as one wastes months.
- If you genuinely cannot fall asleep before 2–4am no matter what, and you have been like this for years, that pattern has a name and a clinical treatment path. Start with why you cannot fall asleep until 3am.
- If you work nights or rotating shifts, the advice above is partly wrong for you, because you are not trying to align with daylight. See sleeping around shift work.
- If you fall asleep fine but wake at 3am and cannot get back, that is a different mechanism from a delayed clock — see why you wake at 3am.
- If you snore heavily, wake unrefreshed however long you sleep, or someone has seen you stop breathing, ask your doctor about sleep apnoea before anything else. No schedule change fixes that.
- If low mood, hopelessness or anxiety are in the picture, sleep and mood move together and treating only one is slow going. Raise both.
And if what is actually waking you is pain rather than timing, the schedule is not your problem — our condition guides cover the night-time setup for specific pain problems.
Frequently asked questions
How long does it take to fix a sleep schedule?
In the randomised trial above, a practical four-part routine shifted night owls about two hours earlier over roughly three weeks, which works out at 30 to 60 minutes a week. A shift of an hour or less can come faster. Anyone promising an overnight fix is describing one bad night of sleep deprivation, not a reset.
What is the single most important thing?
A fixed wake time that you keep at weekends, with light soon after waking. Everything else is secondary, and in the UK Biobank data the consistency of sleep timing predicted mortality more strongly than sleep duration did.
Should I nap while fixing my schedule?
A short early-afternoon nap is a much smaller setback than sleeping in, and it can make holding the wake time survivable. Keep it short and keep it early; a long or late nap removes the sleep pressure you are trying to build.
Does melatonin fix a sleep schedule?
It can help shift it when the dose is low and the timing is right, but it is a supporting tool rather than the mechanism. Morning light and a fixed wake time do the structural work. Most shop-bought melatonin is dosed several times higher than the research uses.
Is it bad to have a different schedule at weekends?
A consistent couple of hours’ difference is what researchers call social jetlag, and a systematic review of 41 studies found it associated with worse health outcomes. If you change one thing, make the weekend wake time match the weekday one.
Sources
Based on articles retrieved from PubMed:
- Windred DP, Burns AC, Lane JM, et al. Sleep regularity is a stronger predictor of mortality risk than sleep duration: A prospective cohort study. Sleep. 2024;47(1):zsad253. DOI
- Facer-Childs ER, Middleton B, Skene DJ, Bagshaw AP. Resetting the late timing of ‘night owls’ has a positive impact on mental health and performance. Sleep Medicine. 2019;60:236–247. DOI
- Chang AM, Aeschbach D, Duffy JF, Czeisler CA. Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. PNAS. 2015;112(4):1232–1237. DOI
- Chaput JP, Dutil C, Featherstone R, et al. Sleep timing, sleep consistency, and health in adults: a systematic review. Applied Physiology, Nutrition, and Metabolism. 2020;45(10 Suppl 2):S232–S247. DOI
- Adhikari P, Pradhan A, Zele AJ, Feigl B. Supplemental light exposure improves sleep architecture in people with type 2 diabetes. Acta Diabetologica. 2021;58(9):1201–1208. DOI
- Photograph by bruce mars on Unsplash.