The sleep regressions guide covers what these disruptions are and what the evidence says about managing them. This article goes a step further back — to the question of whether the regression framework itself holds up.

Search “sleep regression” and you’ll find hundreds of articles telling you exactly when they happen, how long they last, and what to do. The 4-month regression. The 8-month regression. The 12-month. The 18-month. Some lists go further: 3 weeks, 6 weeks, 3 months, 6 months. On and on, right through toddlerhood.

Search the medical literature for “sleep regression,” and you’ll find almost nothing.

That gap — between the confident popular narrative and the near-absence of scientific literature — is the most important thing to understand about sleep regressions. Because the concept has taken on a life completely separate from any evidence base.

What the term means — and doesn’t mean

“Sleep regression” is a popular parenting term. It is not a clinical or medical one. There is no formal definition in sleep medicine. No diagnostic criteria exist. No clinical guidelines are built around it.

The term describes a recognisable experience: a baby who was sleeping reasonably well starts waking more, resisting sleep, or napping poorly. That experience is real. But the term adds something the experience itself doesn’t support — the implication that these disruptions happen at predictable ages, to most babies, in a recognisable pattern.

The word “regression” is also worth examining. It implies going backwards. In reality, the periods popularly labelled as regressions are almost always driven by developmental progression — crawling, language, increased awareness, new cognitive capacities. The baby’s brain is moving forward; sleep is temporarily catching up.

The evidence on predictability

This is where the popular narrative comes apart.

Dr Jodi Mindell is one of the world’s most published researchers in paediatric sleep. She conducted an analysis of survey data collected from thousands of mothers — data from a study published in the journal Sleep — specifically looking for evidence of the predicted regression pattern: consistent spikes in sleep problems at the ages the regression framework predicts.

She didn’t find them.

Her findings, summarised by Dr Craig Canapari — Director of the Yale Pediatric Sleep Center — are striking: only around 28% of parents of 3-month-olds reported increased sleep problems. Only around 30% of parents of 5-month-olds. These are meant to be peak regression ages [1].

That means if you’re told “the 4-month regression is coming,” statistically, more parents won’t experience major disruption than will. The prediction is creating anxiety about something that may never arrive.

The Wonder Weeks controversy

The regression framework is closely tied to the “leaps” concept — the idea, popularised by the Wonder Weeks book and app, that babies experience predictable windows of fussiness and developmental change at specific ages. The leaps model provides the timing framework that underlies much popular regression guidance.

The problem is that the leaps model was scientifically tested and failed.

Carolina de Weerth was a PhD student of Hetty van de Rijt, one of the original Wonder Weeks researchers. She conducted her own study examining whether the predicted fussy periods actually occurred — measuring both observed behaviour and cortisol levels in infants at the ages the leaps model predicted. She found no evidence of the predicted patterns. The predicted fussy windows didn’t materialise in either the behavioural data or the physiological data.

Van de Rijt reportedly attempted to prevent the study’s publication. The controversy that followed led to her departure from her academic position [1].

What is supported by evidence

This doesn’t mean developmental disruptions to sleep don’t exist. They do. It means the rigid, predictable schedule doesn’t.

What the evidence supports:

Developmental milestones can affect sleep. Major motor milestones (rolling, crawling, pulling to stand), cognitive changes (object permanence, language), and emotional shifts (separation anxiety) are all associated with periods of more disrupted sleep in some babies. This is real.

The 4-month sleep architecture change is the real exception. Around 3–4 months, infant sleep permanently matures from newborn sleep biology to adult-like architecture with distinct light, deep, and REM stages. This is a specific, documented biological change with a clear mechanism. It’s the only “regression” that deserves that label in a scientific sense.

Individual variation is enormous. Research consistently shows that infant sleep is highly variable [3]. Some babies sail through the periods that devastate other families. Neither outcome is predictable from outside.

Why the regression narrative persists

If the evidence is this thin, why is the framework so dominant?

It’s commercially useful. Sleep training companies, apps, and consultants benefit from a framework that creates predictable parental anxiety. The regression is coming. You need to prepare. The programme tells you what to do. This is a highly effective marketing structure.

Confirmation bias reinforces it. When parents are told to expect a disruption at 8 months and their baby has a rough week at 8 months, the framework feels validated — even though babies have rough weeks at all ages. The framework gets credit for the hits and is forgiven the misses.

Naming gives a sense of order. Infant sleep is genuinely chaotic and variable. A named, timed framework offers the illusion of predictability and control. For many parents, that framing is genuinely comforting — even if the underlying evidence doesn’t support the precision it implies.

This isn’t to say parents are wrong to find the regression framework useful. For some, knowing “this might happen” is reassuring. The issue is when predictive certainty drives purchasing decisions, creates pre-emptive anxiety about disruptions that never arrive, or — critically — leads parents to believe they failed to prevent something that was never predictable to begin with.

What you should actually expect

Expect variability. Your baby’s sleep will have good patches and bad patches throughout the first two to three years. This is normal and documented [3, 5].

Expect that developmental milestones can disrupt sleep at any age — not only the predicted ones. A baby who has a rough patch at 7 months, or 14 months, is not out of sequence. Sleep disruption doesn’t follow a timetable.

Expect that most disruptions resolve within a few weeks without specific intervention. If a difficult patch persists for many weeks and doesn’t seem tied to a passing developmental phase, that’s worth looking at more carefully.

Don’t pre-worry. Anticipating a regression that may not arrive is a significant source of unnecessary parental anxiety — and it’s being driven by a framework with very thin scientific foundations.

References below.