The 4-month sleep regression: what is really happening and what helps Happyly baby sleep coaching app 4 month regression guide

Around 3 to 4 months, your baby's brain completes a major piece of construction work. Newborn sleep has only two stages — active sleep (similar to REM) and quiet sleep. By 4 months, the brain has built out the full four-stage cycle that adults use: light sleep, deeper light sleep, deep sleep, and REM.

This is a permanent, one-way change. The old two-stage pattern does not come back. That is why parents who describe the regression as 'my baby forgot how to sleep' are actually witnessing the opposite — the baby's brain has learned a more complex way to sleep and has not yet figured out how to transition smoothly between the new stages.

Scott Rivkees's research on infant circadian development, published in Pediatrics, places this transition in the window when the circadian system fully comes online — melatonin rhythms stabilise by about 3 months, and by 4 months the interplay between sleep pressure and the circadian clock (what Alexander Borbely called the two-process model) is running for the first time. The baby now cycles between light and deep sleep roughly every 45 to 60 minutes. At each cycle boundary, there is a brief partial arousal. Adults have these too — we just roll over and fall back asleep without remembering it. Babies have not learned that skill yet.

The practical result: more wake-ups. Not because the baby is sleeping worse, but because the brain is now doing something more complex at every cycle transition.

Parents almost always describe the same pattern: a baby who was sleeping long stretches suddenly wakes every 1 to 2 hours, naps collapse to 30 minutes, and every settling technique that worked last week stops working.

Here is what is typically changing:

The 30-minute nap deserves special attention. It is not a coincidence that naps land at exactly 30 to 45 minutes — that is the length of one sleep cycle at this age. Your baby is completing a full cycle, reaching the partial arousal at the transition, and waking fully instead of rolling into the next cycle. This is the same mechanism driving the night wakings, just more visible during the day.

The reason old settling methods stop working is also biological. When your baby was on the two-stage system, being rocked or fed to drowsiness was enough — they would slip into quiet sleep quickly. Now they enter light sleep first and stay there for 10 to 20 minutes before moving deeper. If the conditions change during that light sleep phase (you put them down, the rocking stops, the breast is gone), they wake. The solution is not to rock harder. It is to give the new system time to mature.

Most families describe the worst of it lasting 2 to 6 weeks. Jacqueline Henderson's research on infant sleep consolidation, published in Sleep Medicine Reviews, found that by 5 months roughly half of infants had returned to sleeping an 8-hour night stretch — but the other half had not, and both groups were developmentally normal.

What does "over" look like? It does not mean your baby goes back to the pre-regression pattern. The two-stage sleep architecture is gone permanently. What it means is that your baby learns to navigate the new four-stage cycles more smoothly — the partial arousals still happen, but your baby gets better at drifting back to sleep through them.

Some honest timelines:

• Weeks 1-2: The worst of the night wakings. Everything feels broken. This is the acute phase.
• Weeks 3-4: Most babies start stringing together slightly longer stretches again (3-4 hours). Naps may still be short.
• Weeks 4-6: A new pattern emerges. It may not look like the old pattern, but it is more predictable.
• Beyond 6 weeks: If sleep has not improved at all after 6 weeks of consistent routine, it is worth talking to your pediatrician — not because something is wrong, but because they can rule out anything medical and discuss next steps.

Marie-Helene Pennestri and colleagues at McGill University published a landmark study in Pediatrics showing that a substantial proportion of 6-month-olds were not sleeping through the night by any definition — and that this predicted neither worse developmental outcomes nor higher rates of maternal depression. The takeaway: do not panic if your baby is not back to long stretches by an arbitrary deadline.

You are reading this at some desperate hour and you need something to try. Here is what actually helps during the regression:

1. Shorten wake windows by 15-20 minutes. During the regression most babies need shorter wake windows, not longer ones. The brain is working overtime on the sleep architecture rebuild and sleep pressure builds faster. If your 4-month-old was doing 2-hour wake windows, try 1 hour 40 minutes. See our wake windows by age guide for the full age-banded chart.

2. Offer an early bedtime. If naps were short all day, pull bedtime earlier by 30-45 minutes. A 6:00 PM bedtime is not unusual during the regression. Overtiredness from accumulated short naps makes the night wakings worse — Marc Weissbluth's clinical observation of the cortisol-driven 'second wind' applies directly here.

3. Try resettling for 10 minutes before declaring a nap over. When your baby wakes at 30 minutes, do not immediately pick them up. Wait a moment. Some babies fuss for 3 to 5 minutes and then drift back into the next cycle. If they do not resettle after 10 minutes, the nap is over — move on.

4. Protect the bedtime routine. Consistency in the pre-sleep routine matters more during the regression than at any other time. Same steps, same order, same room, same voice. The routine is the anchor when everything else is shifting.

5. Use whatever settling method works for your family tonight. A contact nap in a carrier, a feed to sleep, a bounce on the yoga ball — none of these are 'bad habits' during the regression. They are survival tools. You can refine later. Right now, sleep is the priority.

6. Take shifts if you have a partner. The regression breaks parents as much as it breaks sleep. If one caregiver takes the first half of the night (8 PM to 1 AM) and the other takes the second, both get a 4-5 hour unbroken block. That is the difference between coping and not coping.

Short answer: no. Not unless you want to.

The regression will resolve on its own for most families. The baby's brain needs time to learn the new sleep cycle transitions, and it will learn them — with or without formal sleep training. Michael Gradisar's 2016 randomised controlled trial in Pediatrics found no long-term difference between families who used graduated settling, bedtime fading, or no intervention at all. Harriet Hiscock's Australian trials followed families for five years and reported the same: no harm from either gentle intervention or from waiting it out.

Some families do choose to introduce gentle settling strategies during or after the regression. That is a valid choice. Happyly, a baby and toddler sleep coaching app, supports responsive, graduated approaches — not rigid methods. If you want to explore what that looks like, the FAQ on Happyly's coaching approach is a good starting point.

What we would not recommend: making a major sleep training decision in the first two weeks of the regression, when you are at your most exhausted and least able to be consistent. Let the acute phase pass. Then decide.

For a deeper look at the evidence for and against different approaches, see our gentle sleep training guide or the research page for the full citation list.