If you search for it, you will find hundreds of articles about the four-month sleep regression, most of them with a confident explanation and a product to sell. Here is something you will not find in most of them: no peer-reviewed study defines, tests or measures the prevalence of a “sleep regression” at four months, or at any other age. The term does not appear in any of the major systematic reviews of infant sleep.
That does not mean your nights haven’t got harder. It means the popular label is describing something real with the wrong word.
What is actually changing
Between roughly two and five months, a baby’s sleep reorganises in ways that show up clearly on a sleep lab’s instruments.
Newborns fall asleep straight into light, active sleep — the twitchy, noisy kind — which is part of why they transfer so easily from your arms to the crib. Somewhere in this window, babies start entering sleep through deep, quiet sleep instead, the way adults do. That is very likely why “he used to fall asleep in my arms and now the moment I put him down he’s awake” arrives when it does.
At the same time, sleep spindles and other mature features of deep sleep appear on the EEG; the melatonin rhythm, which is essentially absent at birth, establishes itself somewhere between about six and fifteen weeks; and the system that builds up sleep pressure across the day comes online. The body clock is, quite literally, being installed.
Why it feels like going backwards
Here is the part worth holding on to: every population dataset shows sleep consolidating fastest across exactly this window. In the largest pooled review, the longest stretch of night sleep grows by about thirty-nine minutes a month between one and six months — roughly four times the rate it grows in the second half of the year. The biggest single jump in consolidated sleep across the whole first year happens between one and four months.
There is no population-level dip at four months in any normative dataset. What there is, is a change in how your baby falls asleep and in what they notice when they surface between cycles. A baby who used to sink from light sleep into deep sleep without noticing the transfer now enters sleep differently, and a baby who is more awake to the world is more awake to the fact that the conditions they fell asleep in (your arms, the feed, the rocking) have changed.
It is a progression in the physiology, even when it feels like a regression in the house.
What genuinely helps at this age
This is the first age at which any structure is physiologically possible — before about eight to twelve weeks there is nothing to schedule against — so a few things start to earn their keep.
- A short, consistent bedtime routine. The best-supported change in the field, with most of its benefit showing within three nights.
- Bright days and dark nights. The melatonin rhythm is being built; light is the strongest signal.
- Watch the clock at bedtime. Morning wake time becomes remarkably fixed from about five months, so a later bedtime tends to subtract sleep from the front without adding it at the back. In one movement-monitor study, every hour earlier at sleep onset bought about thirty-four extra minutes of night sleep.
- Stop swaddling if you haven’t. Rolling attempts usually begin around three to four months, and swaddling plus a roll onto the tummy is the dangerous combination.
- Drowsy-but-awake, when they’ll tolerate it. A direction to move toward, not a rule you have already broken. Around seventy percent of mothers feed their baby to sleep at this age; it is the norm, not a problem.
What does not help
Any form of sleep training. Extinction-based methods have never been tested in babies under six months — not “tested and found harmful”, simply not tested — and among parents who tried controlled crying without support, the large majority started before six months and reported lower success than they expected. There is a real developmental reason: you cannot schedule against machinery that isn’t installed yet. Foundations now, and revisit at six months.
Where the real dips are
Genuine worsening of sleep is documented in the literature — just later. Newly crawling and newly walking infants have measurably more fragmented sleep, and the pooled night-waking average actually rises between seven and eleven months before falling again. If you want a research-backed “regression”, that is the one. It goes with the milestone, and it passes.
None of this is medical advice. If your baby’s sleep changes suddenly alongside snoring, breathing pauses, feeding difficulty, poor weight gain or clusters of jerking movements on waking, those are questions for your pediatrician, not for an article.