SleepMedic / Guides
Three twelve-hour nights in a row is one of the most common schedules in American healthcare, and one of the least written-about. This is the whole thing: what to do before night one, how to hold the middle, how to get home alive, how to sleep through a Tuesday afternoon, and how to come out the other side without spending your days off recovering from your days off.
The short version
One
Three nights, usually 1900 to 0700 or 1800 to 0600, then four days off. The nights sit back to back, and that is the part that makes the pattern survivable rather than punishing: you are asking your body to hold one schedule for three days instead of flipping it every other day. Inside the block, the work is maintenance. Outside it, the work is transitions, and there are exactly two of them. Going in, you have to be awake and useful at 0300 on a body that has been awake since morning. Coming out, you have to unwind three days of inverted living in about one day so that the rest of your week belongs to you. Everything in this guide hangs on those two hinges. The middle nights are not the hard part. The edges are.
I have worked most of the shapes this job comes in, including this one, so the honest thing first: nobody sleeps as well on three nights as they would on three days. Nothing here fixes that. What it does is make the difference smaller, and make the days off yours instead of a slow crawl back to baseline. The rules underneath are the same handful laid out in the Shift Worker's Field Manual. This page is what they look like pointed at one schedule.
Two
Night one is a physics problem. If you got up at seven in the morning and your shift ends at seven the next morning, you are asking your brain to run twenty-four hours on a system built for sixteen. Sleep pressure — adenosine, accumulating from the moment you wake — does not care that you are at work. (Walker, Huberman Lab Guest Series Ep. 1, 2:36:48) By 0300 you have enough of it in you to fall asleep in a chair. The whole trick of the first night is to arrive owing less of it than you otherwise would.
Pre-pay the pressure. 20 to 30 minutes, or a full 90.
This is the highest-leverage thing on the page, and it is guideline-level advice rather than a trick: the 2025 AASM clinical practice guideline on shift work disorder lists a nap before the shift among its options for the sleepiness side of the problem, alongside bright light on shift and caffeine. A nap taken before you go in is not laziness banked; it is pressure spent early so there is less of it waiting for you at four in the morning.
Keep it short or keep it whole. Twenty to thirty minutes stays shallow enough that you wake clean. Ninety gets you through a full cycle and out the other side. The forty-five-minute nap is the one that hurts, because it pulls you out of deep sleep into an hour of feeling worse than before you lay down.
From the field: Jocko Willink's version is a six-to-eight-minute power nap, feet elevated above the heart. Short enough to stay shallow, and the fixed ritual — same posture, same short duration — is part of why it works. Practice, not a trial.
Where to put the anchor. Everything in a SleepMedic plan is computed backward from one number: your wake time, held per day-type, protected within about thirty minutes. Regularity is king, and the wake time is the strongest signal you can send both clocks. (Walker, Guest Series Ep. 2, 6:53) On a night block that anchor belongs in the mid-afternoon. Waking around three in the afternoon puts an 1800 shift start three hours later, leaves room for the evening nap, and puts your sleep opportunity in the darkest, quietest part of the morning after. Wake at noon and you are running on fumes by 0400. Wake at five and you never get a nap, never get light, and go in cold.
Night one is the exception, because you have not slept during the day yet — you woke up on a normal schedule. That is fine. Sleep in if you can, take the evening nap seriously since it is doing all the work, and let the anchor take hold from the morning of night two onward. The block finds its rhythm on day two.
The first cup. Hold it sixty to ninety minutes after waking rather than immediately. Caffeine blocks adenosine; it does not clear it, and drinking it into the teeth of sleep inertia mostly wastes it. (Walker, Guest Series Ep. 2, 59:25) A mid-afternoon anchor puts the first cup somewhere around four o'clock, which sounds insane to anyone on days and is exactly right for you. Your morning is their afternoon. Stop apologizing for it.
Three
Three things run the middle of a night block: light, caffeine, and the 0300 trough. Two of them you control precisely. The third you plan around.
900 to 6,000 lux, for at least an hour
There is a dose here and most people underdo it badly. Pooling the trials, the effect belongs to sustained brightness: somewhere between 900 and 6,000 lux for at least an hour during the shift did the most work, and a lamp glanced at for ten minutes did almost none (Scientific Reports meta-analysis, 2024). An hour is a long time to sit under a light. That is the finding, and it is why the desk lamp somebody bought you did nothing.
The same analysis carries a caveat worth hearing before you spend money, and for once it is good news for this schedule: light therapy during the shift helps permanent night workers substantially more than it helps rotators. Three nights back to back is not permanent nights, but it is far closer to it than a schedule that flips every other day. Your clock gets to move a little in one direction and stay there for three days. That is the structural advantage of the 3x12 block, and light is how you cash it in.
Front-load it. Get the bright exposure in the first half of the shift and back it off in the last couple of hours, so you are not fighting your own alerting signal on the drive home.
Eight hours before your day-sleep target — usually mid-shift
Caffeine's half-life is about five to six hours on average, faster for some and slower for others depending on a liver-enzyme gene, which is why your partner can have an espresso at dinner and you cannot. (Walker, Guest Series Ep. 2, 59:25) Working backward from that, we cut caffeine eight hours before the sleep target you are protecting.
Compute it from the sleep target, not from the end of the shift. If your day sleep starts at 0700, the last caffeine of any kind lands at 2300 — which is barely a third of the way into a 1900 shift. That is the uncomfortable arithmetic of a 12-hour night, and the thing to notice is that the cutoff arrives hours before the wall does. Fight the wall with light and movement, not with a cup at 0400 that will still be circulating when you lie down.
The 0300 to 0500 trough. Somewhere in there your circadian rhythm bottoms out at the same hour your sleep pressure peaks, and the two systems that normally take turns are both pointing the same direction at once. (Walker, Huberman Lab Guest Series Ep. 1, 2:37:33) That is not a character flaw arriving on schedule; it is the trough, and it is the same two hours for essentially everyone working nights. You will not talk yourself out of it. What you can do is know where it is and refuse to make it worse: do not schedule the hard cognitive task there if you have any say, do not eat a heavy meal into it, and do not treat the way you feel at 0400 as information about how you are doing overall. It passes around dawn whether you deserve it or not.
Move through it. Stand up, walk, go outside for two minutes if outside is available, get under the brightest light you have. Movement and light do more for the trough than anything you can drink, and they cost you nothing at the other end.
20 to 30 minutes, or split the window
If your workplace gives you a real break and somewhere horizontal, take it. Twenty to thirty minutes keeps you shallow enough to wake usable. Anything longer than that and shorter than a full ninety leaves you climbing out of deep sleep into a job.
And if the break structure is broken — which is most places, most nights — split it instead of skipping it. Across a simulated sixteen-hour night, a 90-minute nap plus a later 30-minute nap held alertness as well as a single 120-minute block, and on some measures better (Scientific Reports, 2023). That matters because the two-hour break is the one nobody actually gets. Two smaller windows you can defend beat one perfect window you cannot.
Four
The most dangerous part of a night shift is the drive after it. Put a number on it: when researchers ran nurses through a driving simulator on their real commute after a real shift, the night-shift group crashed on 11 percent of drives against 7 percent for the day-shift group, with more drifting out of lane and more of the hard braking that means you saw it late (James & James, Nursing Reports, 2023). That is a simulator, so nobody got hurt. On the interstate you do not get the version where nobody gets hurt.
If your eyes are heavy at the wheel: pull over and take a 20-minute nap before continuing, use caffeine as an emergency tool knowing it will cost you day sleep, or find another way home. Arriving 40 minutes later beats not arriving.
Sunglasses, and mean it. Dark ones, on before you leave the building, on for the whole drive. Light is the lever that sets the circadian clock, and morning light is the strongest advance signal there is. Sunglasses on the drive home after a night shift are not a style choice; they keep daylight from telling your clock the day is starting just as you need it to end. Keep them on through the parking lot, the front door, and the kitchen. The last hundred feet is where most people take them off, and the kitchen window at 0715 in July is brighter than any lamp you own.
Your entire sleep window, against a world that is awake
Daytime sleep loses to lawnmowers, doorbells, group texts, and guilt. Treat the defenses as equipment, not preferences.
What to eat, and when. This is the part of night-shift advice that changed recently, and it changed in a direction that helps you. Researchers put people through simulated night shifts and changed exactly one variable: whether they ate during their biological daytime or around the clock like a normal night worker. Same calories, same food, same shifts. The daytime-eating group did not develop the internal circadian misalignment the other group did, and their cardiovascular risk markers stayed flat where the others' climbed (Chellappa, Scheer et al., Nature Communications, 2025). Timing was the whole intervention. Not macros, not clean eating, not a supplement. When you eat.
Which is a lovely finding and a hard ask, because a twelve-hour night with no food is not a plan, it is a fantasy. So here is the fallback, and it is better than it sounds: in a crossover trial in night-shift nurses, a single low-glycemic-index meal on shift left glucose profiles close to what they looked like when the nurses ate nothing at all (Soltani et al., J Nutr, 2024). One meal, low GI, early in the shift. Not three, not the vending machine at four. And if you want the version with the most evidence behind it, fasting across the night shift itself has been studied directly and reviewed as a strategy for night workers (review of meal timing and fasting on night shift, 2024) — eat your real meal before you go in and after you get up, and keep the overnight hours as empty as your body will tolerate.
Practically, on a 3x12: a proper meal in the evening before the shift, one modest low-GI meal early in the night, something small on the way home if you are genuinely hungry. Finish eating at least three hours before wind-down starts. Digestion raises core temperature and competes with the thing you are about to try to do.
Five
Here is where most people lose their days off, and they lose them in the first six hours.
You come off night three at 0700 owing a real sleep debt, and every instinct says to sleep until you are done. Do that — go down at eight and get up at four in the afternoon — and you have just handed your clock a full eight-hour day sleep on the day you were supposed to leave nights. That night you lie awake until two in the morning. The next day you are wrecked and off-schedule. One good decision at 0800 has cost you two days.
So the site's rule, and it is the same rule that runs the plan builder: the day you come off a block is a recovery day, not a day off. A recovery day gets a shorter sleep held near your work-block anchor, one planned nap if you need it, and an early night — rather than a full sleep held to your rest-day anchor. Chasing eight hours in the morning is what steals the following night. The rest-day anchor belongs on the clear days after it. Going into the rest of your week rested is decided on the second day off, not the first.
Concretely: sleep a short block after night three, ending mid-afternoon rather than late afternoon. Get bright light the moment you get up, outside if there is any outside available, because that is the signal that drags the clock back toward days. Move. Eat a real dinner at dinner time. Then go to bed early that evening, earlier than feels natural, and let the short morning sleep make that possible. You will not sleep a full night; you will sleep enough, and you will wake on day two roughly where everyone else is. One deliberately imperfect day buys you three real ones.
From the field: I put the alarm somewhere high and across the room, so the only way to silence it is to get out of bed — then the rule is that I do not get back in. It matters most on the recovery day, when the bed is warm and the argument for five more hours is very good. No trial behind this one; the trick is making the return trip harder than staying up.
Name it, because it is the one nobody writes about. After four days off, you have spent most of a week going to bed at a normal hour, and then you walk in at 1900 and have to be awake and competent until 0700. Your clock has drifted back toward days — that is exactly what you asked it to do on the recovery day — and now you are asking it to reverse in a single evening. The first night back is reliably the worst night of the block. Not the third. The first.
There is no version of this that costs nothing. There is a version that costs less: sleep in on the day of night one, take the evening nap seriously, get your light on shift early, and accept that night one is a survival night rather than a performance night. Do not schedule anything the morning after it that requires you to be a person. And do not read the way you feel on night one as evidence the plan is failing — the plan is what makes night two better, which it will be.
Six
Most people never get a vote on this. Read it anyway, because a scheduling committee that has never heard the argument is sometimes one conversation away from making everyone's life easier for free.
If you have any say at all in how the schedule is built, ask for it to move forward — days, then evenings, then nights — rather than backward. Your internal clock drifts a little longer than 24 hours on its own, so a forward rotation asks it to do what it already wants to do, and a backward one fights it every cycle. This is no longer folk wisdom among schedulers. Forward rotation is now guideline-level advice (AASM clinical practice guideline, 2025), and a cohort of 144 nurses working both directions found the forward group less sleepy and measurably better at sustained attention (Lammers-van der Holst et al., JAMA Network Open, 2021). If you sit on a scheduling committee, this is the cheapest intervention on the list. Nobody has to buy anything.
The other structural ask is smaller: keep the nights together. Three consecutive nights is far kinder than three nights scattered across a week, because consecutive nights let the clock settle into one position instead of being asked to move four separate times. If your schedule already does this, that is the good news buried in a hard pattern.
Seven
Some sleep problems are not schedule problems. If you are consistently unable to sleep even with the window protected, if you are falling asleep involuntarily on shift, if you snore hard and wake unrefreshed, or if the dread of not sleeping has become its own thing keeping you awake, that is worth a real evaluation. Shift work sleep disorder is a recognized diagnosis, sleep apnea does not care what shift you work, and the best-evidenced treatment for chronic insomnia is not a pill, it is CBT-I, a structured behavioral program. A sleep clinician can sort out which one you are dealing with.
As of May 2025 that clinician has something they did not have before. The American Academy of Sleep Medicine published its first clinical practice guideline devoted to shift work disorder, which means the advice above stopped being a collection of good studies and became a document a physician can point at (AASM clinical practice guideline, 2025). It suggests bright light during the shift, caffeine before or during it, and a nap before the shift for the sleepiness side; CBT-I for the insomnia side. For the specific problem of not being able to sleep during the day, melatonin taken before daytime sleep is offered as an option, and the honest framing matters: that recommendation is conditional and rests on low-certainty evidence, so treat it as something to try under guidance, not a dose to go buy. And one recommendation runs the other way. The guideline suggests against benzodiazepines for shift work disorder insomnia. If that is what you were handed, that is a conversation to reopen.
One more thing worth knowing, especially if you are a nurse and especially if you are the one who builds the schedule: NIOSH publishes a free training on shift work and long work hours built specifically for nurses and the people who manage them (NIOSH, revised 2023). It is public domain, it is written for exactly this audience, and most units have no idea it exists. It is the document to attach to the email.
This guide is education, not medical advice, and the strongest move available sometimes is asking for help.
Yes, and the 2025 AASM clinical practice guideline on shift work disorder lists a nap before the shift among its recommendations for the sleepiness side of the problem. Going into night one you have usually been awake since morning, which means twelve hours of accumulated sleep pressure arriving right as the shift starts. An evening nap before the first night pre-pays some of that. Keep it short, 20 to 30 minutes, or take a full 90 so you wake at the end of a cycle rather than out of deep sleep.
More than most people get, and for longer. Pooling the trials, the benefit belongs to sustained brightness: somewhere between 900 and 6,000 lux for at least an hour during the shift did the most work, and a lamp glanced at for ten minutes did almost none. The same analysis carries a caveat worth hearing: light therapy on shift helps consistent night workers substantially more than it helps rotators, because a clock that turns over every few days never gets far enough in one direction for the light to hold onto.
Eight hours before the sleep opportunity you are protecting, computed backward from your day-sleep target rather than from the end of the shift. Caffeine has a half-life of roughly five to six hours, so a cup at the end of a night shift is still working on you when you lie down at eight in the morning. On a 12-hour night the cutoff usually lands mid-shift, hours before the wall hits. Plan for that instead of being surprised by it.
Because the commute lands at the worst possible hour, with maximum sleep pressure meeting the bottom of the circadian curve, and because daylight on your face is telling your clock the day is starting. When researchers put nurses through a driving simulator on their real commute after a real shift, the night-shift group crashed on 11 percent of drives against 7 percent for the day-shift group. Wear dark sunglasses home, and if your eyes are heavy, pull over and nap 20 minutes before continuing.
Treat the day you come off the block as a recovery day, not a day off. That means a shorter sleep held near your work-block anchor, ending mid-afternoon, then bright light, then an early night. Chasing a full eight hours on that morning is what steals the following night and turns one transition into three ruined days. The rest-day anchor belongs on the clear days after the recovery day, not on the recovery day itself.
Everything above is the reasoning. The times are the part that has to be yours, because a 1900 start and an 1800 start put the caffeine cutoff an hour apart and the sleep target somewhere else entirely. Three places to get them:
The free plan builder takes your pattern and your two wake times and computes your anchors, light windows, caffeine cutoff, nap window and wind-down for the nights, the recovery day, and the days off. No account, printable, and nothing is sold to you at the end of it.
Build my free sleep planFree, no account
One short email a week for shift workers: what the research actually says, and what to do with it on your schedule.
One email a week. Unsubscribe anytime.