Wake Windows by Age, and How Solid the Numbers Are

Here is the wake windows chart you came for. Here is also something you will not find on most pages that publish one: reputable sources disagree with each other by up to two hours at the same age.

Wake windows are a useful idea. Sleep pressure does build the longer a child is awake, and it builds faster in younger children. That much is real.

What is not real is the precision. There is no normative dataset of infant awake intervals by month, and “wake window” is not a term in the peer-reviewed sleep literature at all. Use the chart as a loose ordering device. If your baby does not fit it, the chart is the thing that is wrong.

Wake Windows by Age

How Much Sleep Babies Need in Total

This is the number with actual consensus behind it. The American Academy of Sleep Medicine convened 13 experts, screened 864 articles, and issued these figures, endorsed by the AAP:¹

That first row is the most interesting thing on this page. The AASM deliberately declined to make a recommendation for infants under 4 months, stating it was due to the wide range of normal variation in duration and patterns of sleep, and insufficient evidence for associations with health outcomes.¹

Hold that alongside the fact that the same age group is routinely given wake windows specified to 15 minute precision. The most cautious body in the field would not state a daily total. The parenting internet states the interval between naps.

How Much Sources Disagree

We compared five widely used wake window charts. The disagreement is not trivial:

• 0 to 4 weeks: published ceilings range from 40 minutes to 90 minutes. That is a 2.25 times spread on the most anxiety-inducing number in the whole category.
• 5 to 6 months: ceilings range from 2.5 to 4 hours, a 60 percent difference.
• 10 to 12 months: ceilings range from 4 to 6 hours. A full two hour disagreement at one age.
• The age bands themselves differ. Some group 14 to 24 months together, others split 15 to 18 and 18 to 24, others publish month by month. The bands are editorial choices, not developmental findings.

Notably, the most medically credentialed source we looked at, Cleveland Clinic, publishes consistently the widest ranges.² That is what honesty about variance looks like, and it is also why their version is the least usable as a schedule. Precision and accuracy pull in opposite directions here.

What Real Schedules Look Like

Charts describe what a schedule should be. Our State of Baby Sleep 2026 report, based on 74,379 parents across 112 countries, describes what they actually are.

Two findings are more useful than any wake window number. First, bedtime is the strongest predictor of wake-up time: 55 percent of babies put down after 9pm sleep past 7am, against 9 percent of those down before 6pm. Second, 1 in 4 babies are up before 6am.

If early rising is your problem, our data says the lever is bedtime, not the interval between naps.

Where the Idea Comes From

Wake windows are a folk translation of something real, so it is worth knowing what the real thing is and what it does not say.

The two-process model

Sleep regulation is usually described with two interacting processes. Process S, homeostatic sleep pressure, builds during wakefulness and dissipates during sleep. Process C is the circadian rhythm. The model is well established and was reappraised by its original authors in 2016.³

But Process S is operationalized as EEG slow-wave activity, which requires polysomnography to measure, and the model was developed on adults and rodents. Nothing in it specifies a wall-clock interval for putting any individual person to bed. The two-process model lends wake window charts an authority it does not actually confer.

Has anyone measured how fast sleep pressure builds by age?

Once, and not in infants. A study found the Process S build-up time constant was 8.9 hours in Tanner stage 1 and 2 children, against 15.4 hours in Tanner stage 5 adolescents.⁴ Younger children accumulate sleep pressure roughly 1.7 times faster, which is the best evidence that the underlying idea is sound.

The youngest participants in that study were around 11 to 14 years old. The build-up rate has never been measured in infants at all.

Newborns have no clock to schedule against

This is the strongest reason to treat newborn wake windows loosely. Circadian rhythms are not present at birth, they develop. Cortisol rhythm emerges around 8 weeks, melatonin rhythm and sleep efficiency around 9 weeks, and core body temperature and clock gene rhythms around 11 weeks.⁵

A separate study found little evidence of rhythmic melatonin excretion before 9 to 12 weeks in term infants, with excretion rising 5 to 6 fold between 6 and 12 weeks. Premature infants showed roughly a 9 week delay, with a 2 to 3 week residual delay even after correcting for gestational age.⁶

So for the first 8 to 12 weeks, Process C is effectively absent. Sleep timing in that window is almost purely homeostatic and feeding-driven. A schedule for a 3 week old has no circadian substrate to attach to.

Overtired and Undertired, Honestly

The conventional signs are worth knowing, and so is which parts of the standard explanation hold up.

The mechanism you usually read is weak

Almost every article on this topic says overtiredness triggers a cortisol and adrenaline spike, producing a second wind. That runs against baseline diurnal cortisol, which peaks after waking and declines across the day to a low point near bedtime. As sleep pressure rises, baseline cortisol is falling.

In newborns the explanation cannot apply at all, because the infant cortisol rhythm does not emerge until around 8 weeks.⁵ Applying an overtired cortisol spike to a 3 week old invokes a rhythm the baby does not yet have. And that is precisely the age group given the tightest windows.

The behavioral observation is fine. Tired infants are more emotionally reactive and harder to settle. The hormonal story bolted onto it is not supported.

A structural problem with “undertired”

Overtired and undertired are claimed to produce overlapping symptoms: short naps, difficulty settling, night waking. There is no independent test to tell them apart. A pair of opposed explanations that jointly cover every outcome and cannot be distinguished from each other is not a testable framework. It is a way of always having an answer.

Cue-based timing is on firmer ground practically, though no study has established the sensitivity or specificity of yawning or eye rubbing for predicting sleep readiness either.

How to Actually Use This

None of the above means wake windows are useless. It means they are a heuristic, and heuristics work better when you know their limits.

• Use the chart as an ordering device, not a target. It tells you roughly when to start watching, not when to act.
• Watch your baby first. Cues outrank the table every time.
• Do not chase 15 minute precision. That precision is manufactured. No dataset supports it.
• Under 12 weeks, hold it loosely. There is no circadian rhythm to schedule against yet.
• Judge by the 24 hour total, which is the number with real consensus behind it.
• If your baby does not fit, that is normal. Reference data shows a 2nd to 98th percentile spread of 10.4 to 18.1 hours of sleep at 6 months of age.⁷

That last point matters more than any row in the chart. With variation that wide, a great many entirely normal babies fit no published table. Content that blames the mismatch on mistimed wake windows converts ordinary population variance into parental guilt, and there is no reason to accept that.

Related reading: sleep regression ages and what is real, when kids stop napping, and what to do when your baby will not sleep in the crib.

Frequently Asked Questions