When Do Babies Sleep Through the Night

The question comes up at every 4 a.m. feeding, every conversation with another sleep-deprived parent, every pediatrician visit where the doctor asks how the baby is sleeping. "When do babies sleep through the night?" It sounds like a reasonable thing to have an answer to. The part nobody mentions is that researchers who study this for a living can't actually agree on what "sleeping through the night" means — five hours, six hours, eight hours, or something else entirely — and that when you look at who sleeps through and who doesn't, the most interesting finding is about cause rather than timing: the two things driving your baby's nights turn out to be completely different problems with completely different levers.

The one-sentence answer: The longest unbroken stretch grows gradually across the first year as the circadian rhythm and sleep architecture mature, but a meaningful number of healthy babies still wake multiple times a night at 6 months and beyond, and a 2025 twin study found that how often your baby wakes and signals is driven mainly by the family's sleep environment, while how intensely your baby cries is mostly genetic. Those two things need different responses.

A quick map of what's below:

• The 2025 twin study that splits night-waking into two separate problems, only one of which parents can reliably change
• Why nights have two acts — and why 4–6 a.m. is hardest
• What breastfeeding does and doesn't do to night sleep, and why
• What happens to night waking without any formal sleep training
• When to mention something to your pediatrician

If the one-sentence answer above is all you needed, you've got the gist. If you want to understand the mechanism behind each piece, keep reading.

Two problems, two drivers: the finding that reframes this completely

Before anything else, it's worth knowing what researchers actually mean when they study "sleeping through the night" — because it's almost certainly not what you think. Different studies use different thresholds: midnight to 5 a.m., any 8 consecutive hours, or a family-realistic 10 p.m. to 6 a.m. window. None uses the 10 to 12 unbroken hours most parenting books imply. A 2010 study by Henderson and colleagues that used videosomnography to check parent sleep diaries found something worth filing away: parents consistently overestimated how long their babies slept, by 43 to 67 minutes a night. "My baby sleeps through" often means a parent slept through, which is not quite the same thing.

That distinction matters because of what a 2025 twin study from Uppsala University clarifies about what's actually happening when babies wake. Viktorsson and colleagues followed 998 Swedish twins at 2 months and again at 5 months, using genetic modeling to separate what drives different aspects of infant sleep.

The finding: the two things parents struggle with most on sleepless nights — how often the baby wakes and signals, and how intensely the baby cries — have almost opposite causes.

The number of night wakeups is driven primarily by the shared family environment. At 2 months, shared environment accounted for 61 to 90% of the variation in how frequently babies woke and signaled. The room setup, the responsiveness patterns, the light exposure, the consistency of the evening sequence (all factors the family controls) are the dominant influence on whether and how often a baby signals across the night.

Crying duration is nearly the opposite. At 2 months, genetics accounted for roughly 50% of variation in how long babies cried. By 5 months, that heritability had risen to around 70%. A baby who cries hard and long when awake at night is largely expressing a trait that comes with the baby, not a problem created by the family's approach.

The practical translation: if you're asking "how do I get my baby to stop waking up so often," that's a question about the environment, and it has evidence-based answers. If you're asking "why does my baby cry for so long when awake at night," that's largely a temperament question, and the honest answer is that it will probably shift as the nervous system matures, not because you found the right technique.

A companion finding from the same study: the ability to settle independently showed a developmental shift between the two ages. At 2 months, settling was primarily environment-driven. By 5 months, genetics had become the stronger influence. The capacity for self-settling develops on a biological timetable — it can be supported by the environment, but it cannot be rushed past the developmental window.

This also means the two most common forms of parental sleep guilt belong to different categories. Guilt about how often the baby wakes is worth examining. Guilt about how hard the baby cries is, to a large degree, misplaced.

Why nights have two acts — and why 4–6 a.m. is hardest

The single number most parenting guides offer ("most babies sleep through by 6 months") describes a median and skips the mechanism entirely. The mechanism is what actually explains your nights.

In the first weeks, there is no circadian rhythm to anchor sleep to nighttime. The pineal gland doesn't produce melatonin rhythmically until around 3 to 4 months, so a newborn's sleep is distributed across the 24-hour day without preference for night. What parents experience as "sleeping in patches" is neurologically accurate — the system that would consolidate night sleep simply isn't online yet.

As the circadian rhythm establishes itself around 3 months, something measurable happens to the longest uninterrupted sleep stretch. A 2025 scoping review by Gilchrist and colleagues in Frontiers in Neuroscience aggregated data from 35 actigraphy studies covering the first 6 months. The developmental trajectory they found: the longest continuous sleep period grows from around 3.6 hours at 3 weeks, to 5.2 hours at 8 weeks, to roughly 6 hours by 12 weeks. That 12-week mark is when the circadian rhythm is coming online — and the two things are not a coincidence.

What this produces in practice is a night with two distinct phases. The first half of the night is dominated by deeper sleep, and that's where the longest unbroken stretch tends to land. The second half, particularly the 4 to 6 a.m. window, is lighter sleep with shorter cycles and more frequent partial arousals. A baby who surfaces at 5 a.m. and won't go back down isn't on a bad schedule — that's the architecture of the developing sleep system expressing itself.

Progress after the 3-month mark continues, but more slowly and less predictably than the first-trimester jump. A 2012 normative meta-analysis by Galland and colleagues in Sleep Medicine Reviews documented how wide the range stays across a healthy population: the spread in sleep patterns is nearly 6 hours wide. Two babies the same age, both developing normally, can have very different nights, and a chart average describes neither of them particularly well.

On baby wake windows, we cover the underlying biology of sleep pressure in more detail, including why the buildup of adenosine during waking, not a clock, is what actually drives when a baby is ready to sleep.

What breastfeeding actually does — and doesn't do — to night sleep

Breastfed babies do wake more often overnight than formula-fed babies. This is consistent across studies, including a systematic review covering more than 6,000 infants. But the follow-on question that most parenting guides skip is: does this mean breastfed babies sleep less overall?

It doesn't. Most studies comparing total sleep find no significant difference between breastfed and formula-fed infants. The distribution is different (breastfed babies tend to have shorter overnight stretches and more distributed waking) but the total volume of sleep is broadly similar.

The biology behind this was laid out in a 2014 paper by McKenna at the University of Notre Dame in the journal Evolution, Medicine, and Public Health. Night waking to nurse serves a function that formula feeding doesn't require: it maintains the frequency of milk removal that supports milk supply, and in the context of evolutionary biology, it extends the inter-birth interval through lactational amenorrhea. Frequent nighttime nursing arousals are a feature of how the breastfeeding system is designed to work, not a sign that it is malfunctioning.

McKenna's research also examined a finding relevant to SIDS biology: infants who died of SIDS appear to have had arousal deficiencies, meaning they did not rouse from sleep readily enough. He argues that practice at arousing — including the many gentle night wakings that breastfed babies do — may function to develop and maintain the arousal capacity that protects against SIDS risk. This is a hypothesis, not a clinical recommendation, and AAP safe-sleep guidelines remain the standard. But it provides biological context for why "getting the breastfed baby to wake less" is not always the straightforward improvement it sounds like.

Breast milk also carries a nighttime melatonin signal. Milk produced at night contains substantially more melatonin than afternoon milk, as documented by Qin and colleagues in Scientific Reports in 2019. This circadian cue in the milk helps calibrate the developing sleep-wake system from the outside, before the baby's own pineal gland is producing melatonin rhythmically — which, as our baby sleep schedule article covers, doesn't begin until 3 to 4 months. Formula doesn't carry this signal.

None of this means breastfeeding families should expect no improvement in overnight sleep. It means the baseline expectation for a breastfed baby should be somewhat different from a formula-fed baby, and that the difference reflects biology rather than something that needs fixing.

What happens without sleep training — and what does sleep training actually change

A common assumption is that sleep training is the mechanism by which babies learn to sleep through the night, and that without it, the problem never resolves. The research is more nuanced than that.

Without any formal intervention, night waking naturally decreases through the first year as the brain matures. The circadian rhythm establishes itself by 3 to 4 months, and the sleep architecture shift that arrives around 4 months is permanent — the pattern before it doesn't come back. Those changes happen on a biological timetable, and a household that doesn't do formal sleep training still benefits from them. Babies who wake frequently at 2 months generally wake less at 6 months, and less again at 12 months, regardless of whether any intervention was applied.

What behavioral sleep training does is accelerate the signaling reduction within that timetable. A randomized controlled trial by Gradisar and colleagues published in Pediatrics in 2016 found that both graduated extinction and bedtime fading reduced night wakings and improved sleep onset compared to control, with no evidence of elevated stress hormones or disrupted attachment at 12 months.

More relevant to the "when does it actually matter" question: a 5-year follow-up of an earlier randomized trial by Price and colleagues, published in Pediatrics in 2012, compared families who had received behavioral sleep intervention to those who hadn't. At 5 years, there were no significant differences between groups in child sleep, emotional health, behavior, or attachment measures. The intervention accelerated the short-term improvement; it didn't produce a different 5-year-old.

The honest framing for families weighing this decision: sleep training is a real tool for reducing signaling frequency in the short and medium term, and the evidence doesn't show that it harms children who receive it. But it is not the only route to a family getting more sleep, and the 5-year follow-up data suggest it does not permanently change the developmental trajectory. A family that chooses not to do formal sleep training is not condemning themselves to five years of broken nights.

The one intervention with the strongest consistent evidence across the literature is a predictable nightly routine, meaning a consistent, sequenced wind-down before sleep. The same underlying mechanism (predictability lowers cortisol, familiarity signals safety) is why routines continue to matter as children get older too, as our article on why daily routines reduce toddler tantrums covers. The environmental setup of the sleep space matters too, which is why calm, low-stimulation nursery environments support the transition to sleep better than stimulating ones. Muted-palette prints work on this principle: the visual environment in the bedroom should feel quiet rather than curious. Our Botanical Nursery Wall Art fits that register, with soft greens and neutrals rather than the bright pattern-rich designs that belong in the daytime play space.

For daytime, active wake windows are where the work of building sleep pressure happens. The developmental activities in our Sensory Play Flashcards 0–12m are built for exactly that window — organized by age and sense, so a two-hour wake window has a clear answer instead of a blank stare at the playmat.

When to mention something to your pediatrician

Normal first-year sleep covers an enormous range. Most sleep patterns that concern parents in the first year (frequent waking, short stretches, irregular naps) fall well within what the research documents as common across healthy, normally developing babies.

A few things are worth flagging at the next well-visit, not because they indicate something is definitely wrong, but because a pediatrician will want to look at them in context:

Any consistent pattern where the baby seems unable to return to a calm state after night waking, especially past 6 months, is the kind of thing a pediatrician will want to know about. So is night waking that appears to be associated with pain (arching, pulling at ears, difficulty settling even with feeding or holding), or a sudden major change in sleep patterns after a period of stability that has no obvious cause like travel or illness. Snoring or labored breathing during sleep is the clearest reason to flag something proactively.

Conversely: waking 2 to 4 times a night at 6 months is not a red flag. Neither is not sleeping through the night by 6 months, or 9 months, or even 12 months. The 2025 Viktorsson twin data, the Galland normative review, and Henderson's longitudinal cohort all document this range as part of normal infant sleep variation.