4-Month Sleep Regression: What It Is + 3 Things to Do

Before about 12 weeks, babies have a two-stage sleep cycle: active sleep (a primitive form of REM) and quiet sleep (a primitive form of deep sleep). Cycles last roughly 50 minutes, and the transitions between stages are smooth — babies generally drift between stages without fully waking.^[1]

Sometime between 12 and 20 weeks, the brain reorganises into the adult sleep architecture: four distinct stages (N1 light, N2 sleep spindles, N3 slow-wave, REM), 90-minute cycles, and — critically — a partial wake-up at the end of every cycle. Adults wake briefly 4–6 times per night and almost never remember it because we re-settle in seconds. Your 4-month-old has the same architecture now but does not yet have the skill to re-settle. So they fully wake, every 45 minutes to 2 hours, all night long.^[2]

This is not a "phase" they "grow out of." This is the new sleep system, permanently. The regression is the gap between the new architecture (here forever) and the re-settling skill (still has to be learned). Most babies close that gap in 2–6 weeks. The ones who close it fastest are usually the ones whose parents understood what was happening and adjusted on purpose.

The 4-month regression has a recognisable signature. Look for three or four of these together, not one in isolation:

The baby was sleeping in longer stretches before this shift, then suddenly is not. A previously 5-hour night stretch shrinks to 90-minute increments. Wakings cluster in the second half of the night (after midnight) — that is the signature of the 90-minute cycle becoming dominant.

The baby resists naps that previously worked, or sleeps for exactly 30–45 minutes and wakes screaming. That is the new cycle length showing up in daytime sleep too.

Bedtime is harder. The same routine that produced a sleeping baby last week now produces a wide-awake baby blinking at the ceiling. Falling asleep takes 30+ minutes when it used to take 5.

The baby is more alert when awake, smiling, tracking, batting at toys. Developmentally something has clearly shifted up — the same brain change driving the sleep regression is also driving cognitive leaps.

If you see only one of those, it is probably something else (teething, growth spurt, illness, a developmental leap that does not affect sleep architecture). If you see three or four, you are in the regression.

Timing-wise: the average onset is week 16 (4 months), but anywhere between week 12 and week 20 is normal. A regression at 3 months is real and means your baby's brain matured early. A regression at 5 months is real and means it matured late. Both are within the normal range.

Everything you learned in the first three months was optimised for a baby with primitive 50-minute sleep cycles and no end-of-cycle wake. After the shift, the same strategies often backfire.

Rocking or feeding to sleep was fine before — the baby went from awake to deep sleep in a smooth slide, and you could put them down in deep sleep with a high success rate. Now there is a light-sleep stage at the start of every cycle. If your baby falls asleep in your arms with motion, then wakes 45 minutes later in their crib with no motion, the brain registers a context mismatch and sounds the alarm. This is why the "transfer to crib" stops working at 4 months even when nothing about your transfer technique has changed.

Strict schedules built around the old wake windows do not match the new ones. From 4 months onward, daytime wake windows extend (about 1.5 to 2 hours) and naps consolidate from many short ones to fewer longer ones. The schedule shape that worked at 12 weeks is too tight by 18 weeks.

Drowsy-but-awake — the technique most parents have heard about and not actually practiced — is now load-bearing. Before 4 months it was a nice-to-have. After 4 months, it is the single most predictive variable for which babies sleep well by 6 months.^[3]

Do not introduce solid food to "fix" the night wakings. The 2022 AAP recommendation is exclusive breast milk or formula until 6 months. Solid food before 6 months does not help night wakings — there is good evidence it does not — and it carries real allergy and gut-development risk. The wakings are not hunger; they are the architecture change.^[4]

Do not abandon the routine because "it is not working." It is working — every consistent wind-down is one rep of the skill. Babies who get an inconsistent wind-down (different every night, sometimes skipped, sometimes 45 minutes long) take longer to come out of the regression than babies who got a flawless 5-minute routine done identically every time.

Do not start co-sleeping during the regression to "get through it" unless you have already chosen co-sleeping as your long-term plan. Starting bed-sharing now and intending to stop later is the hardest pattern to undo. Either commit to it as a long-term choice with the safety guidelines (Safe Sleep 7), or hold the line and stay in your bed.

Do not panic about every 45-minute wake-up. Many of those are partial wakes that the baby will re-settle from in 1–3 minutes if you do not intervene. Pause at the door. Listen. If the noise is escalating, go in. If it is the wind-down sound of a baby falling back asleep, leave them. Many parents create more wakings than the regression does by intervening on every sound.

Most babies move through the 4-month regression in 2–6 weeks once consistent practice begins. The first sign that you are coming out of it is usually a 30-minute reduction in time-to-fall-asleep at bedtime. Then night stretches start lengthening, often jumping from 90 minutes back to 3–4 hours within a few days. Naps consolidate last — that is the slowest piece.

If you are 8 weeks in with no improvement, three things to check before assuming this is your new normal: are the wind-down sequences truly identical and uninterrupted (most parents think they are when they are not — keep a log for a week and review), is your baby actually getting drowsy-but-awake practice or are you putting them down already asleep (this is the most common mistake), and is the bedtime in the right window for the new wake window length.

If wind-down, drowsy-but-awake, and wake-window timing are all dialled in and you are still struggling, the next move is structured sleep training (extinction, gradual extinction, fading, or chair method — choose one). Sleep training is most effective from 4 to 6 months and gets harder after 9 months when separation anxiety begins.^[5] The window you are in now is not just an inconvenience — it is the optimal training window.