How to Sleep Better Working Night Shift Without Fighting Your Body Clock

If you work nights and sleep badly, the first thing to settle is not what time you go to bed. It is whether you are trying to move your body clock onto a night schedule or simply trying to sleep well without moving it. Those two goals need opposite handling of light, and doing one while quietly hoping for the other is the most common reason a night worker's sleep plan does nothing.

The short answer: if you work permanent or long blocks of nights, you can try to shift the clock, using bright light during the shift, dark glasses from the moment you leave the building, and a daytime sleep window at the same clock hours every day. If you rotate, you cannot. The American Academy of Sleep Medicine says so directly in its Management of Shift Work Disorder guideline, dated May 2025: "For individuals who work rotating or frequently changing shifts, circadian adaptation is not a realistic goal." For rotators, the job is damage control done well, which is a different plan with different tools.

The second thing to know is uncomfortable, and no tips list on this subject will tell you. That May 2025 guideline is the first substantive update to shift work disorder guidance in nearly two decades, and it contains 19 recommendations. Every single one is rated conditional. None is strong. Bright light during the night shift, the thing every article leads with, rests on 18 randomised trials and 6 observational studies and was still graded very low certainty. Melatonin for daytime sleep after a night shift was reviewed across 10 trials and the benefit was judged trivial. That does not mean none of it works. It means you are running experiments on yourself with reasonable odds, not following a protocol with guaranteed returns, and you should design your plan so that a failure tells you something.

Your body runs a clock in the brain that sets when you feel sleepy and when you feel awake, and its main input is light hitting the eye. Left alone, that clock holds you awake through the day and drops your alertness in the small hours. A night shift asks you to be sharp exactly when the clock is pushing hardest for sleep, and then to sleep at the hour it is pushing hardest for wakefulness.

That mismatch is why daytime sleep is short and broken even in a dark, quiet room. You are not fighting noise. You are fighting a wake signal.

Light does two separate jobs here, and confusing them is where most plans fall apart. Light makes you feel more alert right now, an effect you get at moderate intensity. Light also moves the clock, an effect that needs more intensity and is entirely about timing. An evidence summary hosted on the NIH's PubMed Central puts the practical thresholds at roughly 1,000 to 5,000 lux for reducing drowsiness without necessarily shifting the clock, and above 5,000 lux for actual phase shifting. Lux is a measure of how much light reaches your eye: an ordinary lit office sits in the hundreds, an overcast day outdoors is thousands. Trials aimed at circadian adaptation in the AASM review used 2,500 to 5,500 lux.

The timing rule is the part people miss. Bright light in the earlier part of your night pushes the clock later, which is the direction a night worker wants. Bright light at the end of the night and on the drive home does the opposite, dragging the clock back toward a day schedule and undoing the work. That is why the guideline's adaptation recommendation is not just light. It is light during the shift plus reduced-light-transmittance glasses, meaning dark or blue-filtering lenses, worn from the end of the shift until you get into bed.

Melatonin is the other signal. Your body releases it when the clock thinks it is night; taken as a supplement, it acts as a mild sleep promoter and a timing cue. The AASM suggests it in three separate situations, all conditionally, and the evidence is thinner than the supplement aisle implies.

What follows is a full cycle, from the afternoon before your first night to the morning you go back to sleeping at night. Run it for one whole block of shifts before judging it. A single night tells you nothing, because sleep debt from the previous week is still in the system.

Step 1. Pick your strategy from your roster, not from your preference. Look at the next eight weeks. If it is permanent nights or blocks of four or more consecutive nights with predictable timing, you are in the adaptation column. If nights, evenings and days are shuffled week to week, you are in the defence column and should stop trying to hold a night-oriented clock. Cost: 10 minutes with the roster. You will know it worked when you stop doing contradictory things, like chasing bright light at 6am on a Tuesday and sleeping at midnight on Wednesday.

Step 2. Bank sleep before the block starts. Extra sleep before an anticipated shift improves subsequent work performance and reduces fatigue, according to the NIH evidence summary. In practice this means going to bed earlier the night before your first night, and adding a nap on that day. Cost: an early night and an hour or two of a free afternoon.

Step 3. Take a nap before the shift. The AASM suggests a nap prior to the night shift over no nap, based on 3 studies, very low certainty. There are two shapes it can take. A short nap of 30 minutes or less improves alertness while keeping sleep inertia, the groggy stretch after waking, to a minimum. Or a longer pre-shift sleep opportunity of 3 to 4 hours, which is the design used in the melatonin trials the guideline reviewed. Choose one and keep it consistent. You will know it worked when the hardest part of the shift, usually the small hours, feels merely tiring rather than dangerous.

Step 4. If you nap, pair it with caffeine on waking. The guideline suggests nap plus caffeine over neither, based on one RCT run in both lab and field settings, on the reasoning that caffeine reduces post-nap sleep inertia. Give yourself at least 20 minutes after waking before driving or anything else that punishes a lapse. The AASM's own wording: "Sleep inertia occurring soon after the nap may temporarily increase sleepiness, decrease cognitive performance, and increase accident risk."

Step 5. Get bright light early in the shift, not late. Bright light during the night shift is suggested for excessive sleepiness on the basis of 18 RCTs and 6 observational studies, rated very low certainty, and the guideline notes it may also help insomnia and daytime function. Put the exposure in the first half of your shift. Sitting in a brightly lit area for a stretch, or using a light box while you do paperwork, is enough; you do not need to stare at anything. Cost: free if your workplace is already bright, otherwise the price of a light box. You will know it worked when your slump arrives later in the shift than it used to.

Step 6. Set your caffeine cutoff and hold it. Caffeine for sleepiness on night shift is suggested on the basis of 8 RCTs, very low certainty. It works. It also has a long tail. Six hours before your sleep window is the cutoff, which for a sleep window starting at 8am means nothing after 2am. If your shift ends at 7am, that means your last coffee comes well before the end of the shift, which is exactly the part people refuse to do and then blame the blackout blind.

Step 7. Put the glasses on before you leave the building. Not in the car park. The whole point is that the daylight between the exit door and your bed is the strongest signal your clock will receive all day, and it points the wrong way. The NIH evidence summary recommends blue-light-filtering glasses, dark goggles or sunglasses for the commute home; the AASM includes reduced-light-transmittance glasses worn from shift end until sleep in its circadian adaptation recommendation. Cost: one pair of glasses, and remembering them.

Step 8. Decide on melatonin with your eyes open. The guideline suggests it in three places, all conditionally. Before a pre-shift nap: 4 RCTs, with doses across those trials ranging from 0.1 mg to 40 mg, taken 30 minutes to 2 hours before a 3 to 4 hour daytime sleep opportunity. For daytime sleep after a night shift: 10 RCTs, very low certainty, benefit judged trivial. For the transition from daytime back to night-time sleep: 3 RCTs, low certainty, with potentially meaningful gains in both self-reported and objectively measured sleep duration. That last one is the strongest of the three, which is worth noticing, because it is the use case nobody markets. The guideline does not identify an optimal dose, and the 400-fold spread across trials is a sign the question has not been answered rather than a menu to pick from. If you try it, start at the low end of what has been studied and change one thing at a time.

Step 9. Protect a 9 to 11 hour sleep opportunity, not a 7 hour one. Opportunity is not the same as sleep. Daytime sleep is broken, so you need more time in the window to land the same total. Block it in the calendar, tell the household, and treat a delivery driver at the door the way you would treat one at 3am.

Step 10. Plan the exit. After the last night of a block, you have to get back to night sleep. The NIH evidence summary recommends at least two nights of recovery sleep between shift rotations, and shifts that never exceed 24 hours. This is where the melatonin evidence is at its best and where most people improvise worst, typically by staying up for 24 hours to force a reset and then sleeping through the following night's alarm.

Three conditions do most of the work, and the NIH evidence summary names all three: dark, quiet, and cool, at around 65°F or 18°C.

Dark means dark at 1pm in July, which ordinary curtains do not deliver. Test it by standing in the room at midday with the door shut and waiting two minutes for your eyes to adjust. If you can read the time on your phone screen reflected off the ceiling, or see a bright outline around the window, the room is not finished. Blackout blinds fitted inside the recess, a blackout curtain that overlaps the frame on all four sides, or a well-fitting sleep mask are the fixes, roughly in that order of effectiveness.

Quiet is harder in daylight because the noise is different in kind: doorbells, deliveries, kids, traffic that peaks at exactly the hours you are asleep. Masking noise with a steady sound is easier than blocking it, since a constant hum raises the floor so individual bangs stand out less.

Cool matters because the same wake signal that keeps you awake during the day also runs your body temperature up. A room at the cooler end helps you fall back asleep after the inevitable mid-afternoon waking, which for most daytime sleepers is the part of the block that actually breaks.

One caution specific to shift workers: the NIH evidence summary advises avoiding alcohol within 3 hours of bedtime. A drink after a night shift feels like it speeds sleep onset, and it does, but it costs you the second half of a sleep block that was already going to be short.

Bright light at the end of the shift. You feel the alertness benefit, so you chase it into the final hours and then drive home in daylight without glasses. Both exposures pull the clock toward a day schedule. How to recognise it: you are doing everything on the list, feel fine at 6am, and cannot fall asleep at 9am. Fix: front-load the light, and put the glasses on indoors.

A nap long enough to reach deep sleep, right before something dangerous. The guideline is explicit that sleep inertia after a nap may temporarily increase sleepiness, reduce cognitive performance and raise accident risk, and advises at least 20 minutes of recovery before safety-sensitive tasks. How to recognise it: you wake up worse than you went down, and it lasts. Fix: cap the nap at 30 minutes or less, or plan for a longer sleep with a real buffer afterward, and add caffeine on waking.

Caffeine doing its job too well. Eight trials support it for alertness on shift. The same guideline notes that "caffeine intake close to bedtime can disrupt sleep onset and quality, which may exacerbate symptoms." How to recognise it: you sleep for two hours after the shift and then lie awake. Fix: move the cutoff to 6 hours before the sleep window and hold it for a full block before deciding it did not help.

Expecting melatonin to fix the daytime sleep block. Across 10 RCTs, the AASM judged the benefit for daytime sleep after a night shift trivial. That is the specific use most people buy it for. How to recognise it: you have been taking it for months and cannot say what changed. Fix: either move it to the use with the better evidence, the transition back to night sleep, or stop and spend the attention on light timing and the room.

Trusting the label on the bottle. The guideline flags that over-the-counter melatonin content has been measured from 83% below to 478% above the stated dose. How to recognise it: you cannot, which is the problem. The same nominal 3 mg from two brands may not be the same dose. Fix: stay with one product once you find one that seems to work, and treat any dose comparison across brands as meaningless.

Reaching for a benzodiazepine. The AASM reviewed 4 RCTs on triazolam and recommends against benzodiazepines for this population, judging fall risk, abuse potential, dependence and withdrawal to outweigh the benefit. How to recognise it: a prescription that started as a short-term bridge through one rough block and is still running. Fix: a conversation with the prescriber about alternatives, which in the guideline include ramelteon at 8 mg (1 RCT) and suvorexant, a dual orexin receptor antagonist, at 10 to 20 mg (1 RCT). Both conditional, both thin.

Running an adaptation plan on a rotating roster. This is the expensive one, because it costs months. If your schedule changes every week, the light and glasses protocol cannot land, and the guideline says adaptation is not a realistic goal in that situation. How to recognise it: you keep starting the plan over. Fix: switch to the defence column, where the aim is total sleep across 24 hours, not a shifted clock.

Which way your roster rotates. The AASM suggests clockwise rotation over counterclockwise. Clockwise means the sequence runs days, then evenings, then nights, which asks your clock to move later each time. Later is the direction the human clock finds easier, the same reason a Friday late night is painless and a Monday early alarm is not. If you have any say in how your unit's roster is built, this is the single change that affects everyone on it rather than one person.

Shift length does not settle it. On 8-hour against 12-hour shifts, the guideline found no preference, based on 5 non-randomised studies rated very low certainty. If someone tells you 12s are ruining your sleep, or that 12s are better because of the longer breaks, the evidence does not currently support either claim.

The transition back to a normal week. This is the one time the melatonin evidence points somewhere useful: 3 RCTs, low certainty, with potentially meaningful gains in sleep duration during the daytime-to-night-time transition. Pair it with at least two nights of recovery sleep between rotations, per the NIH evidence summary.

When it is a diagnosable disorder. The AASM guideline estimates shift work disorder affects about 27% of adult shift workers, with a confidence interval of 21% to 33%, meaning the true figure is likely somewhere in that band. The NIH evidence summary puts the figure at roughly 10% meeting full diagnostic criteria, with 25% to 30% of shift workers reporting insomnia symptoms; the gap between those estimates comes down to the definition used. The practical line is persistence and consequence: sleepiness or insomnia tied to the schedule, lasting months, affecting work or safety. That is a clinical assessment, and it changes what is on the table.

Prescription territory. For excessive sleepiness, the guideline suggests armodafinil and modafinil, wake-promoting drugs, and this is the only place in the whole document where the certainty rises to moderate: 5 RCTs plus 2 observational studies for armodafinil, 3 RCTs for modafinil, with headache and nausea as common side effects. For insomnia symptoms, it suggests cognitive behavioural therapy for insomnia, modified for shift workers, which is a structured programme retraining the association between bed and sleep. None of that is a self-directed fix, and the wake-promoting drugs require a diagnosis. It is worth knowing they exist before concluding that nothing works.

Hours that no plan survives. The NIH evidence summary's workplace recommendations include shifts not exceeding 24 hours, and at least two nights of recovery sleep between shift blocks. If your roster breaches those, the limiting factor is the roster.

Conditional, in the GRADE system the AASM used, means the panel thinks most informed people would choose the option but a substantial number would reasonably not. Very low certainty means the true effect could be substantially different from what the studies found. Put those together and you get an honest picture: these are sensible bets with real mechanisms behind them, tested in studies that were mostly small, mostly short, and often run in laboratory simulations of shift work rather than on actual hospital floors and loading docks.

That is worth understanding before you spend money. The guideline's structure was confirmed in the peer-reviewed literature through an abstract in the journal SLEEP documenting the work, led by Kenneth Wright at the University of Colorado. AASM also ran a public comment period on the draft, open from August 4 to September 1, 2025, inviting clinicians, patients, insurers and advocacy groups to weigh in before finalisation. This is a document built carefully on a shallow evidence base, not a careless one.

What would settle it: larger randomised trials run on real rosters rather than simulations, measuring objective sleep with actigraphy or polysomnography rather than questionnaires, following workers over months instead of days, and reporting outcomes that matter, meaning accidents, error rates and long-term health, not just how sleepy people say they felt at 4am. Until that exists, the sensible approach is the cheap interventions first, in the order the mechanism supports: light timing and the glasses, then the sleep window and the room, then caffeine timing, then supplements last.

The cost ranking helps here. Fixing the timing of light you already receive costs nothing. Glasses and a blackout blind are a one-off. Melatonin is a recurring cost attached to the thinnest evidence of the three, and the labelling problem means you may not be buying what you think you are.