The 4-Month Sleep Regression: What It Is and How Long It Lasts
Share
The acute disruption usually lasts two to six weeks. The underlying change does not end. At around four months a baby's sleep permanently reorganises into adult-like cycles with a full arousal between each one, so things improve when your baby learns to resettle — not when a phase passes.
Last updated: 1 August 2026 · Written by Atikur Rahman
This is one age in a wider pattern. For the others, see the full sleep regression calendar or the free printable chart.
How long does the 4-month sleep regression last?
Two to six weeks for the acute phase. The table below is the usual arc.
| Phase | Timing | What it looks like |
|---|---|---|
| Onset | Usually 3.5–5 months | Waking every 45 minutes to 2 hours. Naps shortening to a single cycle. |
| Peak | Weeks 1–2 | Several night wakings, hard to resettle, 30–45 minute naps. |
| Settling | Weeks 2–6 | Longer stretches return as resettling improves. Naps lengthen last. |
| Past 6 weeks | — | Usually a schedule or habit issue rather than the change itself. |
If nothing has improved at all by around six weeks, stop waiting it out and look at the timing of the day and what happens at each wake-up. That is almost always where the answer is.
What is actually happening at four months?
Newborn sleep alternates between two immature states. At around three to four months it matures into distinct lighter and deeper stages, cycling roughly every 45 minutes, with a brief full arousal at the end of each cycle. Adults do this too — we surface several times a night, roll over and go back down without remembering it.
A baby who has always been put down already asleep now surfaces into a room that has changed. The feed has finished, the arms are gone, the motion has stopped. There is nothing wrong with them; they simply have no practised route back to sleep.
Total sleep need does not drop. The American Academy of Sleep Medicine, endorsed by the AAP, puts infants aged 4 to 12 months at 12 to 16 hours per 24 hours including naps.
Why it does not simply end
This is the part most articles skip, and it is the reason waiting it out often fails. The other rough patches — 8, 12, 18 months — are driven by temporary things: a new skill being rehearsed, a nap schedule that no longer fits, teeth. Those pass on their own.
The four-month change is structural and permanent. Your baby will keep having full arousals between cycles for the rest of their life. What improves is their ability to get back down through one. That is a skill, and it develops either because you give it room to or because it happens gradually anyway — but it does not arrive by the calendar.
What the data actually shows
Worth knowing, because it changes how much you should worry. Large datasets do not show a clear four-month cliff. A 2016 study using sleep-tracking app data found nighttime sleep sessions lengthening steadily through the first year with no dip at four months. A 2018 review found night wakings decreasing across the first year on average — parents report more wakings between three and six months, but wearables do not confirm it.
Day-to-day variation is also larger than most parents expect: nighttime sleep can differ by around two hours from one night to the next in the same baby.
None of that means your experience is imaginary. The change to sleep architecture is real and well documented. It means the severity and timing vary enormously, plenty of babies pass through it without much drama, and a bad week is not evidence that something has broken.
What it looks like
- Waking every 45 minutes to 2 hours through the night, often at predictable times
- Naps collapsing to 30–45 minutes — exactly one cycle
- Needing the same conditions to fall back asleep as they had at bedtime
- Fighting bedtime, or falling asleep easily and waking 45 minutes later
- Being genuinely hungrier — a growth period often overlaps this window
What actually helps
Give them a chance to link cycles before you go in. A brief arousal is not the same as being awake. Waiting a couple of minutes lets some babies close the gap themselves. Going in at the first sound teaches that the arousal always ends with you.
Put them down drowsy rather than asleep, at least once a day. The bedtime settle is usually the easiest one to practise on because sleep pressure is highest.
Fix the timing before anything else. At four months most babies manage roughly 90 minutes to 2 hours awake between sleeps. An overtired baby wakes more, not less. Our wake windows by age chart has the ranges, and there is a free printable version.
Keep the routine identical. Same order, same room, same words. Predictability is doing more work than any single technique.
Do not change the system on night three. If a rough week turns into co-sleeping, hourly feeding or an hour of rocking, that is what persists afterwards. It may be fine with you — but decide it deliberately rather than at 3am.
Feed if they are hungry. Four months is a common growth period. This is not the moment to be rigid about night feeds.
The safety change most people miss
Four months is also roughly when babies start rolling, and that changes two things.
Stop swaddling as soon as there is any sign of rolling — attempting it, not achieving it. A swaddled baby who rolls to their front cannot push up or turn their head freely. Move to a sleeping bag or a swaddle with the arms out.
Keep following AAP safe sleep guidance: baby on their back, on a firm flat surface, in their own clear sleep space, with no pillows, bumpers, loose blankets or soft toys. Room-sharing without bed-sharing is recommended for at least the first six months. Once a baby can roll both ways independently you do not need to reposition them, but they should still be placed down on their back.
When to speak to a clinician
Milestone timing explains a lot, but not everything. Contact your paediatrician or health visitor if disrupted sleep comes with:
- Snoring, mouth breathing, gasping or pauses in breathing during sleep
- Fever, ear-pulling, or waking with a distinct pained cry
- Feeding refusal, persistent vomiting, or a drop off the growth curve
- Loss of a skill your baby previously had
- No improvement at all after six weeks
Frequently asked questions
How long does the 4-month sleep regression last?
The acute disruption typically lasts two to six weeks. The change to sleep architecture underneath it is permanent, so sleep improves as your baby learns to resettle rather than when the phase ends.
Can the 4-month sleep regression start at 3 months?
Yes. The shift in sleep cycles happens somewhere between roughly three and five months, so an early start is normal and not a sign of anything wrong.
Does every baby have a 4-month sleep regression?
No. Large-scale sleep data shows no consistent dip at four months across babies. The underlying change is universal; how much visible disruption it causes is not.
Should I sleep train at 4 months?
Many programmes suggest waiting until around four to six months, and starting during illness or a major leap is harder than waiting. Practising falling asleep in the cot at bedtime is a gentler first step that does not require a formal method.
Why are naps only 30 to 45 minutes?
That is one full sleep cycle. Short naps are the daytime version of the same change, and they usually take longer to resolve than night sleep does.
Should I stop swaddling during the 4-month regression?
Stop as soon as your baby shows any sign of trying to roll, regardless of what it does to sleep. A swaddled baby who rolls onto their front cannot reliably free their arms or turn their head.
Where to go from here
The Wake Windows Guide covers the timing system this sits on top of, and the Gentle Sleep Training Guide sets out the no-cry-it-out approaches for when things have settled. Still in the early weeks? The Newborn Guide covers the run-up to this point.
Free printables: the sleep regression chart and the wake windows chart. Or browse all baby sleep guides.
Sources
- American Academy of Sleep Medicine — Recommended Amount of Sleep for Pediatric Populations (endorsed by the AAP)
- American Academy of Pediatrics — HealthyChildren.org, Infant Sleep
- Stanford Medicine Children's Health — Infant Sleep
- Journal of Sleep Research (2016) — app-based infant sleep patterns
- Journal of Sleep Research (2018) — review of night waking across the first year
- ParentData, Emily Oster PhD — Are Sleep Regressions Real?
Educational only, and not a substitute for advice from your paediatrician or health visitor. Follow current AAP safe sleep guidance.