At 12 weeks, Owen was sleeping 9-hour stretches. We had read the books, run the routine, executed the plan, and we were absolutely insufferable about it. We told other new parents that the secret was just being consistent. We were the Bad People you encounter in parenting forums.
At 16 weeks, Owen started waking up every 90 minutes. Every. Ninety. Minutes. For two and a half weeks. The schedule we’d built shattered. The crib was suddenly a place of personal betrayal. We Googled “is my baby broken.” We were no longer insufferable. We were ruined.
This was the 4-month sleep regression. It is real. It is biological. It has nothing to do with anything you did wrong, and almost nothing you do can prevent it. But it has an end, and there are a few things that meaningfully shorten how rough it is. Here’s what we learned.
What’s actually happening at 4 months
“Sleep regression” is misleading because it implies a temporary setback to a previous state. What’s really happening is a permanent change in how your baby sleeps — they’re moving from newborn sleep architecture to something close to adult sleep architecture, and the transition is rough.
The specifics, from Why We Sleep by Matthew Walker and the AAP’s pediatric sleep literature:
- Newborn sleep has only two stages: Active Sleep (the baby version of REM) and Quiet Sleep. Cycles last about 50–60 minutes. Babies enter sleep through Active Sleep, which is why they can fall asleep on you and stay asleep through a transfer.
- Around 12–16 weeks, sleep matures into 4 distinct stages including light NREM, deep NREM, and REM, with brief arousals between cycles. Cycles still last about 50–60 minutes.
- The catch: at every cycle transition, there’s a brief partial wake-up. Adults do this too — we just roll over and go back to sleep without remembering. Babies don’t yet know how to do that “rolling over” part. They wake up fully and need help.
This is why sleep gets worse at 4 months even though everything else (alertness, motor skills, social engagement) is improving. They’ve gained adult sleep cycles without yet gaining adult self-soothing skills.
How long it lasts
The regression itself — the part where everything is genuinely worse — typically lasts 2 to 6 weeks. The new normal that follows is permanent: from then on, your baby’s sleep is architecturally adult, and the only question is how good they get at managing the cycle transitions on their own.
Some kids self-soothe through transitions naturally within a few weeks of the regression starting. Others need help — which is where the sleep-training conversation begins, and where we ended up.
What actually moved the needle (in order of impact)
1. Drop the swaddle (the week you see the first roll attempt)
Once a baby starts trying to roll, the swaddle goes from “calming the Moro reflex” to “trapping their arms in a position that prevents them from clearing their face if they end up prone.” The AAP guidance is unambiguous on this — discontinue swaddling at the first sign of rolling, no exceptions.
For us, this caused a few horrible nights of the startle reflex waking him up constantly. The transition tools that helped:
- A sleep sack with arms out (Halo, Kyte, or the Merlin Magic Sleepsuit if you don’t mind looking like you’ve packaged your baby like a marshmallow).
- Two nights of swaddle with one arm out, then two nights with both arms out, then full sleep sack. A graduated transition, not cold turkey.
2. Build a self-soothing window into the routine
This is the controversial one, and it’s the one that worked for us. We started putting Owen down drowsy but awake instead of fully asleep. The first few nights he protested. Then his protests got shorter. Then they stopped.
The mechanism: babies who learn to fall asleep at the start of the night have a much easier time falling back asleep at the cycle transitions in the middle of the night, because falling asleep is the same skill in both cases. Babies who only know how to fall asleep being rocked / fed / held wake up fully every 60 minutes and need to be re-rocked / re-fed / re-held to start the next cycle.
You don’t have to do “extinction” cry-it-out to get this benefit. The Ferber method (graduated check-ins), the chair method (sit nearby and gradually move farther), and “the shuffle” all work. The 2012 RCT in Pediatrics by Hiscock et al. showed that all three reduced infant sleep problems and maternal depression with no negative effect on attachment, infant cortisol, or behavior at 5-year follow-up.
The guilt-friendly framing that helped us: every minute they spend learning to fall asleep on their own is a minute of better sleep for them, every night, for the rest of their childhood. We are not depriving them of comfort. We’re handing them a tool.
3. Cap the wake windows
At 4 months, the optimal awake stretch between naps is about 90 minutes to 2 hours. A baby who’s been awake for 3 hours is overtired, and an overtired baby fights sleep harder, sleeps lighter, and wakes more.
The fix is unsexy: watch the clock more than you watch the baby’s tired cues. Tired cues at this age are subtle and easy to miss. Wake windows are reliable. We set a phone timer when he woke up and started naptime preparations 15 minutes before the window closed.
What didn’t work for us
- “Dream feeding” at 11 PM. Some families swear by it. For us, it just woke Owen up and added a feeding back into the night that he was otherwise ready to drop.
- White noise machines louder than a quiet conversation. We had ours cranked at first. Backed it down to ~50dB and slept better ourselves.
- Staying in the room “just to be there.” He was more alert with us in the room and took longer to settle. Counterintuitive but consistent.
- Adding a bedtime bottle “to top him off.” Made him more uncomfortable, not less. The hunger story wasn’t the right story for our kid at this age.
What’s normal vs. what to ask the pediatrician about
The 4-month regression is normal, but it shouldn’t last forever. Talk to your pediatrician if:
- The regression goes longer than 6 weeks with no improvement.
- Your baby is snoring loudly or pausing breathing during sleep (possible sleep apnea — uncommon but worth ruling out).
- Your baby is consistently waking in pain (arching back, screaming, settling only when upright — possible reflux).
- You’re at the edge yourself — the parent is part of the system, and parental sleep deprivation has real medical consequences. Postpartum depression risk is meaningfully elevated by sleep loss. Talk to someone.
The honest closing
We sleep-trained Owen at 17 weeks using a modified Ferber. By night 5 he was sleeping 11 hours straight. By night 10, he was protesting for less than 2 minutes before falling asleep. He’s now a 14-month-old who walks himself into his crib, lies down, and waves at us. Most nights.
I’m not telling you to do what we did. I’m telling you that the 4-month wall is a real biological event, that the playbook for getting through it is well-established and well-studied, and that the version of this you read on Instagram is going to scare you more than the actual research will.
You will sleep again. Probably soon. The wall is real, but it’s not a permanent address.
Related reading: Baby Sleep: A Survival Guide Based on Actual Research covers the broader sleep landscape from newborn through 12 months. Twelve Hours’ Sleep by Twelve Weeks Old is the structured method that worked for us before the regression. The Happiest Baby on the Block covers the 5 S’s for the first three months.