First Son

My First Son did not sleep through the night until he was fourteen months old.

For over a year, he woke two, three, sometimes four times a night. We rocked him. We fed him. We bounced on a yoga ball in the dark at 3am like it was a reasonable thing for adults to be doing. We tried white noise. We tried no white noise. We tried driving him around the block at midnight because someone on the internet said it worked.

My wife was the sole bedtime parent. The sole 1am parent. The sole 4am parent. By month four she was nursing him to sleep every night — it was the only thing that worked — and then lowering him into the crib like she was defusing a bomb. Sometimes it held. Often it didn't, and twenty minutes later he was up again.

I thought something was wrong with him. Turns out, nothing was wrong with him. His brain had just upgraded. And we had spent four months building the exact habits that made that upgrade as painful as possible.

That's the 4-month sleep regression. Not a regression at all. A maturation. And if you know it's coming, you can prepare for it. I didn't know. You can.

What Actually Happened to His Brain

Newborns cycle between light and deep sleep in a simple pattern. They can often drift between cycles without fully waking. Around 3–5 months, the brain reorganizes — permanently — into adult-style sleep architecture. Distinct stages: light sleep, deep sleep, REM. Brief natural awakenings between every cycle.

Those cycles last roughly 45 minutes. When a baby surfaces between cycles, they check whether conditions match. If they fell asleep in a crib, the crib is still there. Nothing to report. They drift back under.

If they fell asleep in your arms — or at the breast, or rocking, or bouncing on a yoga ball — and they wake at 1am in a still, dark crib, the conditions don't match. Something is wrong. They alert. They cry. They need you to recreate the falling-asleep conditions before they can sleep again.

Every 45 minutes. All night.

This is the core insight. The way a baby falls asleep at bedtime determines whether they can fall back asleep between cycles. That's it. That's the whole mechanism.

It's also the mistake I made with First Son for over a year.

Not a regression — a permanent change

Other sleep disruptions at 8 months, 12 months, and 18 months are temporary — caused by developmental leaps that pass. The 4-month shift is different: the underlying neurological change is permanent. Newborn sleep architecture doesn't come back. This is why "waiting it out" doesn't work the way it does at other ages.

Signs You're In It

The Swaddle Warning

The 4-month regression hits exactly when rolling begins. That combination matters.

A swaddled baby who rolls to their stomach cannot push themselves back up or turn their face away from the mattress. Arms pinned. This is a suffocation risk. The AAP is unambiguous: stop swaddling the moment your baby shows any rolling signs — rocking side to side, getting one shoulder off the mat during tummy time, any attempt at a roll.

⚠️ Stop swaddling tonight if any rolling signs are present

This is not a gradual transition. The day you see a rolling attempt is the last night of swaddling. Sleep will get worse for a few nights during the switch to a sleep sack. That's the trade for safety. Make it.

What Made It Worse (First Son)

I read zero books on infant sleep before First Son arrived. I googled things at 2am, absorbed contradictory advice from strangers, and received a single sentence from our pediatrician: "Some babies just take longer." I took that as permission to keep doing what we were doing.

Three things I was doing that made it worse:

Building sleep associations I couldn't sustain. Rocking, nursing, bouncing — all of them became the condition for sleep. Once sleep architecture matured at 4 months and the inter-cycle awakenings multiplied, he needed those conditions recreated multiple times a night. We'd built the requirement. Now we were paying for it at 1am, 3am, and 5am.

Ignoring wake windows. At 4 months, a baby can comfortably stay awake for 1.5–2 hours between sleeps. Go longer and cortisol spikes — the stress response makes it harder, not easier, to fall asleep. I was keeping him up longer hoping he'd sleep better. The opposite happened. The overtired baby who fights sleep isn't undertired. They're over the edge.

Inconsistency. Some nights we rocked. Some nights we tried putting him down awake and lasted four minutes before caving. He had no stable expectation for what sleep looked like. The brain needs consistency to build a pattern. We gave it chaos.

What I Did Differently with Second Son

By the time Second Son arrived, I'd read the research. Not blogs. The actual studies. I understood what was coming at 4 months before it arrived.

I addressed sleep associations early

Starting around 6 weeks, I practiced drowsy but awake — placing him down when sleepy but not yet fully asleep. Not every time. Not even most times. But once a day, I tried it. The goal wasn't perfection. It was introducing the concept. By 4 months, he had some experience falling asleep without being held. That foundation made everything easier when the architecture shift hit.

I picked a method before the regression hit

With First Son, I didn't choose a method. I had a vague sense that I was supposed to "let him cry" or "not let him cry" and toggled between the two based on how guilty I felt. With Second Son, I chose graduated extinction — the Ferber method — before night one of sleep training. A 2016 randomized trial by Gradisar found it helped babies fall asleep faster, lowered cortisol levels, and produced no differences in attachment at twelve months. When night two was harder than night one, I didn't panic. I'd read that this was normal. The plan held because I'd made the decision before I was standing in the dark at midnight.

I watched the clock, not just the baby

At 4 months: 1.5–2 hours awake between sleeps. At 5 months: closer to 2–2.5 hours. By the time a baby is rubbing their eyes and fussing, they're already past the edge. I put Second Son down by the clock. Not when he looked tired. When the window said it was time.

I built the environment and kept it consistent

Blackout dark. Light suppresses melatonin. Not dim — dark. A completely dark room makes a measurable difference in sleep onset and early morning waking.

White noise. Continuous, consistent sound placed between the baby and the door. It masks the household noise that triggers light-sleep-phase waking. Not nature sounds. Not music. A steady, consistent sound.

Same routine, same order, every night. Bath, feed, song, dark room, down. Under 30 minutes. It's a neurological cue, not just a ritual. The brain learns to expect sleep when the sequence runs.

The Part Nobody Tells You

The 4-month mark is not when sleep falls apart.

It's when the habits you built — or didn't build — in the first four months become visible.

If your baby has been falling asleep independently, even sometimes, the transition is manageable. If they've been falling asleep exclusively with rocking, feeding, or bouncing, this is the moment the wheels come off. Not because something went wrong. Because the new sleep architecture creates more inter-cycle awakenings, and every awakening exposes the mismatch.

First Son was fine. He got through it. But we were on that yoga ball at 3am until he was fourteen months old. That didn't need to happen.

Second Son was sleeping through the night by six months.

The difference wasn't the baby. It was knowing what was coming.

The most common mistake

Waiting it out. The underlying sleep architecture change is permanent — this isn't a leap that resolves on its own. Waiting for it to pass the way other regressions pass can turn 6 weeks of disruption into 6 months. The window where sleep associations are easiest to change is 4–6 months, before object permanence and separation anxiety make the process harder.

The second mistake: introducing new sleep associations to survive the regression. Nursing or rocking more during the hard nights digs the hole deeper. I know. I did it.

Know what's coming before it arrives

Scout sends you a monthly email on your baby's birthday covering exactly which developmental windows are opening that month — including sleep changes. The 4-month regression. The 8-month separation anxiety spike. The nap transitions. All of it, timed to your baby's exact age.

Try Scout Free →