Sleep Regression Ages: What Is Real and What Is Not

The commonly cited sleep regression ages are 4, 6, 8 to 10, 12, 15, 18 and 24 months. Here is the part almost nobody tells you: only one of them is well established in the research.

“Sleep regression” is a parenting term, not a clinical one. It does not appear in the International Classification of Sleep Disorders or in any diagnostic manual. That does not mean your baby’s sleep did not fall apart at 8 months. It means the framework you have been handed to explain it is looser than it sounds.

This guide gives you the by-age reference, then tells you honestly what sits behind each one. That second part is what will actually help at 3am.

Sleep Regression Ages at a Glance

The 4 Month Change Is Real, and It Is Not a Regression

This one deserves its reputation, but the name is backwards. Nothing regresses. Your baby’s sleep permanently matures into something more adult-like, and the side effect is that it gets lighter.

Before roughly 2 to 3 months, infant sleep is not even scored the way adult sleep is. It is classified as active, quiet, or indeterminate rather than in the N1, N2, N3 and REM stages used for adults. Sleep spindles, a marker of more mature sleep, emerge around 6 to 9 weeks. K-complexes appear from about 5 to 6 months.

Once cycling matures, your baby surfaces more often into lighter sleep, and each surfacing is an opportunity to wake fully. This does not reverse. It is how they will sleep from now on.

A number worth correcting

You will very often read that infant sleep cycles are 45 minutes. Recent actigraphy work across 152 infants and more than 35,000 hours of sleep put the infant cycle at about 62 minutes, against 81 minutes for their parents, and found cycles lengthen by roughly 10 minutes between 3 and 12 months.¹ The 45 minute figure appears to be folklore.

There is also a finding that cuts against the whole 4 month regression narrative. A study tracking 75 infants with detailed diaries across a full year found the largest single gain in self-regulated sleep, 504 minutes, happened between 1 and 4 months.² The period framed as a collapse is, on average, the period of fastest improvement.

What Happens at the Later Ages

The later regressions map onto real development. The problem is the calendar, not the underlying idea.

Separation anxiety does not arrive in discrete windows

Separation anxiety begins around 8 months, peaks across roughly 10 to 18 months, and resolves by about 24 months. That is one continuous 10 month plateau. The popular framing splits it into three separate 4 week events at 8, 15 and 18 months, which is not how it behaves.

Nap transitions are gradual, over months

The shift from two naps to one does cluster significantly around 18 months.³ But it happens over weeks to months of inconsistency, not as a sudden event. Reference data on 493 children found 100 percent still napping at 12 months, 96.4 percent at 18 months, 50.4 percent at 3 years and 8 percent at 5.³

The 12 and 15 month ones have the least behind them

No study links vocabulary spurts to measured sleep disruption. Walking is real and can disrupt sleep. But the claim that there is a distinct, predictable 12 month regression rests on nothing published.

The Wonder Weeks lineage did not replicate

Much of the leap and regression calendar traces to the Wonder Weeks framework, which proposed 10 strictly timed periods of instability. An independent attempt at ethological validation reported that the results failed to support the 10 period pattern. Worth being fair about the scale on both sides: the replication attempt followed 4 mother and infant pairs, and the original work involved around 15 infants. Neither is a large study.

What the Large Cohorts Actually Show

If regressions happened to most babies at predictable ages, you would see spikes in night waking at those ages across large samples. Three datasets say otherwise.

The largest is a study of more than 1,200 infants measured at 6, 15, 24 and 36 months. Growth mixture modeling found two trajectories rather than universal regressions. About 66 percent were flat sleepers, essentially unchanged from 6 to 36 months. The other 34 percent declined steadily, from about 7 waking nights a week at 6 months to 2 a week at 15 months to 1 by 24 months.⁴

No spikes at 8 months. No spikes at 12. For two thirds of infants, no notable change at all.

A second cohort of 388 infants found 37.6 percent were not sleeping 6 consecutive hours at 6 months, and 27.9 percent still were not at 12 months. Critically, it found no association between uninterrupted sleep and Bayley developmental scores, and none with maternal mood.⁵

And the normal range is enormous. A meta-analysis of 34 studies put mean infant sleep at 12.8 hours with a range of 9.7 to 15.9.⁶ Reference data at 6 months of age shows a 2nd to 98th percentile spread of 10.4 to 18.1 hours.³ That is a 7.7 hour spread at a single age.

Why nobody can settle this

A 2025 scoping review of actigraphy and diary studies in the first 6 months found that 28 of 35 studies reported two or fewer age data points. Almost nobody samples densely enough in time to detect a 3 week event. So the honest position is that the 4 month regression cannot be firmly confirmed or refuted by current data, and the later ones have never been properly tested at all.

Where Night Waking Actually Peaks in Our Data

The regression calendar puts the headline event at 4 months. Our own survey of 74,379 parents across 112 countries in the State of Baby Sleep 2026 report shows the peak somewhere else entirely.

Night waking peaks at 7 to 9 months, when 66.7 percent of babies wake three or more times. It rises through 4 to 6 months rather than spiking there, and it declines gradually afterward rather than resolving in discrete 4 week windows.

That shape matches the longitudinal research described above, and it does not match the regression calendar. What parents experience as a series of separate events looks, across tens of thousands of families, like one long curve.

What Actually Helps

Short list, because the evidence base is short.

A consistent bedtime routine

This has randomized trial support. Across 405 mother and child pairs, introducing a consistent routine improved sleep onset latency, longest continuous sleep, and the number and duration of night wakings, and improved maternal mood.⁷ A follow-up found more routine nights per week produced better outcomes.

Reduce what has to be recreated at 3am

This is the one piece of advice with a formal basis. The clinical category that most persistent “regressions” actually fall into is sleep-onset association type insomnia: the child needs the same conditions recreated after every normal arousal. Bedtime problems and frequent night waking affect 20 to 30 percent of infants and young children.⁸ If your baby falls asleep being fed or rocked, they will look for that at each surfacing.

Keep safe sleep intact

Worth stating because regression content drives tired parents toward inclined sleepers, positioners, and sofa co-sleeping. None of those are safe. See our guide to what to do when your baby will not sleep in the crib for the safe options at 3am.

What does not help

There is no evidence that any specific regression responds to any specific intervention, and none for the widely repeated claim that inconsistency makes a regression last longer. If someone tells you your 8 month regression became permanent because you were inconsistent, they are making that up.

When It Is Not a Regression

The biggest practical risk of the regression frame is that it explains away things that need attention. Raise these with your pediatrician rather than waiting them out:

• Snoring, mouth breathing, or pauses in breathing during sleep
• Poor weight gain
• Ear pain, or waking that started with an illness and did not resolve
• Sudden onset in a child whose sleep had been genuinely robust
• Pain-like waking: arching, inconsolable crying, distress after feeds
• Symptoms on 3 or more nights a week for 3 months or more, with daytime impairment, which is the formal threshold for chronic insomnia⁸

For related reading, see our guides to wake windows by age, when kids stop napping, and the Ferber method and how it compares.

Frequently Asked Questions