When Do Babies Sleep Through the Night? What the Research Actually Shows

The definition problem, stated plainly

Almost every disagreement about this question dissolves once you pin down the criterion. Consider a single hypothetical infant who reliably sleeps from 7pm to 1am, feeds, then sleeps from 1:30am to 6:30am.

Under the classic midnight-to-5am criterion, this baby fails — there is a feed at 1am. Under a six-consecutive-hours criterion, this baby passes easily, twice over. Under an eight-consecutive-hours criterion, the baby fails. Under a 10pm-to-6am criterion, the baby fails. Under the colloquial parent definition of "sleeps through," most people would say no.

One baby, one night, five different answers. Now scale that across a study population and across the literature, and the reason published ages range from around three months to well past twelve becomes obvious. The infants are not that different. The rulers are.

There is a second layer to this. The midnight-to-5am definition, which produces the earliest and most reassuring ages, was not designed to describe a baby's experience. It was designed around the parent's — five hours in the middle of the night is roughly the block of sleep an adult needs to function. It is a measure of parental relief, not of infant sleep consolidation. That is a perfectly legitimate thing to measure, but it answers a different question from the one most parents think they are asking.

The practical takeaway is a defensive one. Whenever you encounter a specific age attached to this milestone — in an article, a book, a comment thread, or from a well-meaning relative — treat it as incomplete until you know the criterion behind it. An unqualified number in this domain carries almost no information.

The definitions used in the literature, side by side

The table below sets out the criteria that appear most often in published infant sleep research, what each one actually requires, and the direction it pushes the reported age. It is worth reading as a translation key: when you next see a claim about this milestone, you can usually work out which row it came from.

Note the last column especially. Criteria are not neutral. A study is not simply measuring when babies sleep through the night — it is measuring when babies meet a threshold someone chose, and the choice of threshold determines the headline finding more than the infants do.

Henderson 2010: the study that made the problem visible

The clearest demonstration of the definitional problem comes from a longitudinal study by Henderson and colleagues, published in Pediatrics in 2010 under the title "Sleeping through the night: the consolidation of self-regulated sleep across the first year of life." Rather than picking one definition, the researchers followed a cohort of infants across the first year and assessed each of them against three criteria simultaneously: five consecutive hours, eight consecutive hours, and the fixed 10pm-to-6am block.

The design is what makes it valuable. Because the same infants were scored three ways, the differences between the resulting curves cannot be attributed to different samples, different countries, different feeding practices, or different measurement tools. They are attributable to the criterion alone.

What the study found, in shape rather than in decimal places: the three curves separate substantially. The age at which half the sample qualified moved progressively later as the criterion got stricter, with the fixed 10pm-to-6am window — the one closest to what parents typically mean — producing the latest milestone of the three. And notably, even at twelve months, a meaningful minority of infants had not achieved the strictest criterion. Sleeping through the night, on any demanding definition, is not a universal first-year accomplishment. The published paper reports the month-by-month percentages for each criterion, and it is worth looking at the actual figures rather than taking anyone's summary of them, including this one.

The second contribution of the paper is in its subtitle: "the consolidation of self-regulated sleep." The framing there is that the developmental achievement is not sleeping without waking — infants surface between sleep cycles throughout the night at every age, as does everyone reading this — but returning to sleep afterwards without needing an adult to reinstate the conditions. That reframing matters, because it means the milestone is about self-settling rather than about the absence of arousal, which is a physiologically impossible standard.

Pennestri 2018 and the question nobody had checked

The second study worth knowing is "Uninterrupted Infant Sleep, Development, and Maternal Mood," published by Pennestri and colleagues in Pediatrics in 2018. It approached the topic from an unusual angle: rather than asking when babies achieve consolidated night sleep, it asked whether it matters developmentally if they do not.

The study assessed a cohort of infants at six and twelve months against both a six-hour and an eight-hour uninterrupted criterion, and separately measured mental and psychomotor development and maternal mood.

The prevalence findings alone are useful context. At six months, fewer than half of the infants slept eight consecutive hours, while a clear majority managed six. Both figures rose by twelve months, but even at a year, uninterrupted eight-hour sleep was far from universal. In other words, the baby who is still waking at six months is squarely inside the normal distribution rather than at its edge.

The more consequential finding concerned outcomes. The study found no significant association between uninterrupted sleep and the infants' measured mental or psychomotor development. Babies who slept in longer blocks did not score better on developmental assessment than those who did not. What the study did find was an association with maternal mood — mothers of infants sleeping in longer uninterrupted blocks reported better mood.

That pairing is the honest version of this whole topic. The evidence does not support the idea that a baby who wakes at night is being developmentally harmed by it. It does support the idea that the parent is having a much harder time, which is a real problem deserving a real response — but it is a different problem from the one usually stated. Framing night waking as an infant deficit to be corrected misplaces where the cost is actually falling.

Two trajectories, not one staircase

A third strand of evidence complicates the intuitive picture of steady progress. Weinraub and colleagues, publishing in Developmental Psychology in 2012, tracked infants' nighttime awakenings from six through thirty-six months in a large cohort and looked for patterns of change over time.

Rather than a single population curve, the analysis identified distinct trajectories. One group — the majority — showed the expected pattern: waking declining steadily through the second half of the first year and largely resolving. A second group, roughly a third of the sample, continued waking regularly, with the frequency declining much more slowly and persisting well past the point at which their peers had stopped.

This is a different picture from the standard one, and a more forgiving one. The standard picture implies a single track that all infants travel at different speeds; if yours is slow, they are behind. The trajectory finding implies something closer to distinct patterns, where a persistently waking eighteen-month-old is not a delayed version of a consolidated one but is following a different and reasonably common course.

The study also examined what predicted membership in the persistently waking group. Factors associated with it included infant sex, breastfeeding, and aspects of temperament and maternal characteristics — that is, largely constitutional and contextual factors rather than anything resembling parental error. Note that association is not causation and cohort studies of this kind cannot establish direction; breastfeeding, in particular, is associated with more frequent night waking in multiple datasets without that constituting an argument against it.

The practical value of knowing this is mainly protective. If your baby is at twelve or eighteen months and still waking, the literature contains a well-documented group that looks exactly like that. It is not evidence that something went wrong.

What the normative reviews say about waking rates

Beyond individual studies, there are systematic reviews that pool observational data to establish what is normal at each age. The most widely used in this area is Galland and colleagues' "Normal sleep patterns in infants and children: a systematic review of observational studies," published in Sleep Medicine Reviews in 2012.

The review's most useful contribution is not any single number but the size of the variation around every number. At every age examined, the spread between infants at the lower and upper ends of normal is large — large enough that two babies of identical age can differ by hours of total sleep and several night wakings while both sitting comfortably inside the normal range.

This matters for how you read any statistic about infant sleep. Averages in this domain describe a population, not a target. A baby sleeping considerably less than the mean for their age is not necessarily sleep-deprived, and a baby waking more often than the mean is not necessarily disordered. The reference ranges exist to flag genuine outliers, not to grade individuals.

The same principle applies to the total-sleep recommendations. The American Academy of Sleep Medicine's pediatric consensus recommends 12 to 16 hours per 24 hours for infants aged 4 to 12 months, including naps, and 11 to 14 hours for children aged 1 to 2 years. Those ranges are four and three hours wide respectively, which is the consensus panel's way of acknowledging that the honest answer varies enormously between individuals. Note also what the consensus does not cover: it makes no recommendation for infants under four months, explicitly because the evidence and the variation at that age do not support one. Anyone giving you a firm sleep-duration target for a newborn is going beyond what the AASM was willing to state.

Why parent reports and objective measurement disagree

There is a systematic gap between what parents report about night waking and what objective measurement — actigraphy or video — records, and understanding it resolves a lot of confusion.

The gap runs in a consistent direction: objective measures detect more awakenings than parents report. This is not carelessness. A parent can only report the wakings they are notified about. An infant who surfaces at the end of a sleep cycle, stirs, and returns to sleep without vocalising has genuinely woken, and a video or actigraph will record it, but nobody in the household will ever know.

This produces a useful reframing. When a baby "starts sleeping through the night," what has usually changed is not that they stopped waking. Arousals between sleep cycles continue throughout life. What has changed is that they stopped signalling — they now resettle themselves rather than calling for the conditions that were present when they fell asleep to be restored. Researchers sometimes distinguish "signallers" from "self-soothers" for exactly this reason, and the distinction explains why the milestone can appear to arrive abruptly.

It also explains a pattern that otherwise looks like backsliding. A baby who was self-resettling and starts signalling again after an illness, a trip, or a developmental leap has not lost the underlying capacity. The conditions changed, and the signalling resumed. Once it settles, the previous pattern usually returns without any intervention.

One further implication: because the reported milestone depends on signalling rather than on sleep architecture, it is partly a function of the household as well as the infant. Whether a stir escalates into a full waking is influenced by proximity, monitor sensitivity, and how quickly an adult responds. None of that is a criticism of any particular arrangement — it is just a reason the same baby can "sleep through" in one setup and not another.

Where the twelve-hour expectation came from

If the research criteria top out at eight hours, why do so many parents expect twelve?

Part of it is arithmetic. A baby who goes down at 7pm and wakes at 7am has slept twelve hours, and once a family lands on an early bedtime — which is common and generally sensible from the second half of the first year — twelve hours becomes the span the night is measured against. The expectation is a byproduct of the schedule rather than a claim about infant physiology.

Part of it is commercial. The twelve-hour night is a strong promise, and it appears in the marketing of books, programmes, courses, and consultants far more often than it appears in peer-reviewed work. It is not a fabrication — plenty of babies do sleep twelve hours, and the AASM range of 12 to 16 hours per 24 hours is easily compatible with a twelve-hour night plus naps in the second half of the first year. But the frequency with which it is presented as the expected outcome is out of proportion to the evidence that it is typical.

And part of it is social. Twelve hours is the version that gets reported at playgroup, partly because it is the version worth mentioning. Nobody volunteers that their eleven-month-old still wakes at 2am. The reporting is filtered, and the filtered version becomes the perceived norm.

None of this means a twelve-hour night is unreasonable to want or unrealistic to reach. Many families do, particularly after the first year. It means that measuring your baby against it in month five, on the basis of research that never tested it, is measuring against a standard the evidence does not actually establish.

What the guidance bodies do and do not say

It is worth noting what the major clinical bodies actually publish on this question, because their restraint is informative.

The American Academy of Pediatrics, through its HealthyChildren materials, describes typical infant sleep patterns and the general trend toward longer night sleep across the first year, and offers guidance on routines and safe sleep. What it does not do is issue a target age by which a baby should be sleeping through the night, or define the phrase as a milestone to be met.

The American Academy of Sleep Medicine's pediatric consensus statement recommends total sleep amounts per 24 hours by age band. It says nothing about how that sleep should be distributed, and makes no recommendation at all below four months of age.

The UK's NHS guidance on helping a baby to sleep is similarly framed around variation and practical routine rather than around milestone ages, and is explicit that babies differ widely in when they begin sleeping for longer stretches.

The absence is the point. If there were a reliable age at which infants should achieve consolidated night sleep, these bodies would be well placed to state it, and they have not. The confident ages circulating in parenting content do not originate from clinical consensus.

Separately, and importantly: whatever your baby's sleep pattern, safe sleep guidance is not variable. The AAP recommends infants sleep on their back, on a firm flat surface, in their own space cleared of soft bedding, and room-sharing without bed-sharing for at least the first six months. That guidance is worth reading in full at the source — the AAP's safe sleep hub and the NICHD's Safe to Sleep campaign are both linked below. This site does not attempt to summarise it, because it is not something to take second-hand.

Using the evidence without it making you feel worse

The reason to know all of this is not academic. It is that the confident, unqualified ages circulating in parenting content cause real distress to people who are already depleted, and the evidence provides some genuine relief from them.

Four things follow from the research above. First, any age you are given is meaningless without its criterion, so a baby who is "behind" on one definition may be ahead on another — and the definition was chosen by a researcher, not by nature. Second, the developmental evidence does not support the idea that a night-waking infant is being harmed; Pennestri and colleagues looked directly for that and did not find it. Third, persistent waking well past the first birthday is a documented and reasonably common trajectory rather than a failure. Fourth, the parental cost is real and separately worth addressing — the same study that found no infant developmental effect did find an association with maternal mood, and that is not a small finding to leave lying on the table.

That last point deserves emphasis, because the honest conclusion of this evidence is not "relax, it does not matter." It is that the burden is landing on you rather than on your baby, and that is a legitimate reason to change something — whether that means adjusting the schedule, sharing nights differently, considering one of the behavioural approaches covered on our page on how to sleep train a baby, or getting help. Chronic parental sleep deprivation is a health issue in its own right, and persistent low mood after birth is worth raising with a clinician rather than absorbing.

If what you actually wanted was the practical picture — what to reasonably expect at each age and what tends to shift when — the companion page on when do babies start sleeping through the night is organised as a developmental timeline rather than an evidence review. And the homepage has the broader schedule-by-age picture that this question usually sits inside.