The 4-month sleep regression — what it actually is and how to fix it
Your baby was sleeping. Now they wake every 45 minutes. This is not a phase. Here is what happened and what to do.
At around 3.5–4 months, your baby's sleep architecture permanently changes. Before this point, newborn sleep was relatively simple — a short cycle of active sleep followed by quiet sleep, and back again. After this point, sleep becomes adult-like: multiple stages, more frequent surfacing between cycles, lighter overall.
The problem is not the change itself. The problem is what happens when your baby surfaces between cycles at 3am and the thing that got them to sleep — breast, bottle, dummy, rocking — is no longer there.
Most content about the 4-month regression describes it as a temporary phase that passes. This is wrong, and believing it is why many families are still managing frequent night waking at 8, 10, and 12 months. The architecture change is permanent. What resolves is the disruption — but only once the baby learns to resettle between cycles independently.
What is actually happening in your baby's brain
Before 4 months, your baby's sleep looked like this: active sleep → quiet sleep → active sleep → quiet sleep. Short cycles, simple pattern, brief surfacings that produced no real waking.
After 4 months, sleep looks like this: light sleep → deep sleep → REM → light sleep → repeat. This is the adult sleep architecture. The cycle is still approximately 45 minutes long, but each surface into light sleep is now a genuine partial waking.
In an adult, this is imperceptible — we drift back into the next cycle without noticing. In a baby who has learned to fall asleep with a prop (breast, bottle, dummy, rocking arm), each surface produces a full wake-up because the prop that created the original sleep association is no longer present.
The baby is not broken. The baby is not manipulating you. The baby is doing exactly what any person would do if they fell asleep with a pillow and woke to find it gone — looking for it.
The only resolution is teaching the baby to fall asleep without the prop at the start of the night. When they can do that, they can resettle the same way between cycles. The night waking stops because the problem at 3am is the same problem as the one at 7pm — and when 7pm is solved, 3am follows.
What most parents do that makes it worse
Waiting for it to pass
The most common and most costly mistake. Because the 4-month regression is described as a phase everywhere online, most parents wait it out. But the architecture change is permanent — there is nothing to wait for. A baby with an unaddressed settling association at 4 months will still have it at 8 months, 12 months, and beyond. The waking frequency may reduce slightly as the baby matures, but the pattern does not resolve on its own.
Introducing new settling props during the regression
The regression hits, the baby stops sleeping, the parent starts rocking or feeding to sleep because it works. This solves the immediate problem and creates a longer-term one. The new prop becomes the new association. Now instead of a dummy association, there is a rocking association. Instead of a feed association, there is a bouncing-on-a-gym-ball association. The regression ends but the night waking continues because the association continues.
Blaming teeth, growth spurts, or illness
Teething does not cause persistent nightly waking across weeks. Neither do growth spurts (which last 2–3 days, not 3 months). Illness temporarily disrupts sleep but does not permanently change the settling pattern. If the waking has been consistent for more than 2 weeks without a clear medical cause, it is almost certainly the 4-month regression — the sleep architecture change, not the tooth coming in.
Fixing naps before fixing nights
Short naps at 4 months are caused by the same mechanism as night waking — the baby surfaces between cycles and cannot link them. Some parents focus intensively on nap fixing while the overnight settling association remains. This is backward. Fix overnight settling first. When the baby learns to fall asleep independently at night, nap linking typically follows within 2–3 weeks without any direct nap intervention.
What to expect, night by night
- Night 1–2
Typically the hardest nights. The baby is confused by the change in settling conditions and protests significantly. Most parents see 40–60 minutes of protest on night 1, reducing to 20–30 minutes on night 2. This is the extinction burst beginning.
- Night 3–4
Protest often reduces significantly. Many families see the first extended stretch of sleep (3–5 hours) during this period. This is the association beginning to change — the baby is learning that settling occurs independently.
- Night 5–7
Most babies show consistent improvement by night 5–7 with a consistent approach. Night waking frequency reduces, self-settling between cycles improves. Some babies require a full 10–14 nights for complete consolidation.
- Week 2–3
For babies with deeply entrenched associations (those who have been rock-to-sleep or fed-to-sleep for many weeks), the consolidation period is longer. Consistent response is the key variable — a single night of reverting to the old association typically requires 2–3 additional nights to recover.
What actually works
Move the bedtime feed earlier in the routine
The single highest-impact change. If the last thing that happens before the baby goes into the cot is feeding, feeding is the sleep association. Move the feed to before the final activity (bath, book, song) so the baby is drowsy but awake when placed in the cot. This one change resolves a significant proportion of 4-month night waking within 5–7 nights without any further intervention.
Use a consistent goodbye phrase
A short phrase said the same way at the end of every settling — "night night, sleep tight", "I love you, sleep well", whatever feels natural — gives the baby a reliable signal that the interaction is ending. Said once, warmly, then leave. Over 5–7 nights the phrase itself becomes a settling cue. The baby begins to associate it with the onset of sleep rather than with the departure of a prop.
Respond consistently overnight
When the baby wakes between cycles overnight, the response should be the same as the bedtime settling — brief, calm, the goodbye phrase, and leave. The specific method matters less than the consistency. A response that is different on night 1 and night 3, or different between parents, teaches the baby that the response is unpredictable — which increases rather than decreases the calling out.
Check the schedule before changing settling
A baby who is overtired will fight sleep regardless of settling method. At 4 months, wake windows should be 75–120 minutes. A baby put down after only 60 minutes of wake time is unlikely to settle easily. Before assuming the settling approach is not working, confirm the schedule is appropriate for the age.
Three things to do tonight
- 1Move the bedtime feed before the last book or song — not after. The baby should be awake when placed in the cot.
- 2Choose a goodbye phrase and use it once at the end of settling. Say it the same way every time from tonight.
- 3When the baby wakes overnight, wait 3 minutes before going in. Then: phrase once, leave. Same response every time, both parents.
Related sleep schedules
Build the routine before the next regression hits.
The 4-month regression is the optimal moment to build a sleep routine. The circadian rhythm is established, the baby responds to consistency, and associations are not yet deeply entrenched. A structured approach now produces results in days.
Build my routine — $45 →Related guides
Questions about the 4-month sleep regression
Methodology grounded in paediatric sleep research and evidence-based clinical practice. Schedule data, wake window recommendations, and regression timelines are derived from published guidelines including those of the American Academy of Sleep Medicine (AASM), the National Sleep Foundation, and paediatric sleep research published in peer-reviewed journals.
Read our full methodology →