Sleep Regression Chart: Every Age, What Changes, How Long

How to read the chart

Four things before the table, because a chart read the wrong way does more harm than good.

The ages are approximate and they overlap. A baby may hit the eight-to-ten-month cluster at seven months or at eleven, and more than one entry frequently runs at once — a ten-month-old can be rehearsing a new motor skill, working through separation awareness and outgrowing a nap in the same fortnight. Untangling which is which is most of the work.

The fourth column is the point. "Documented" means the underlying biological change has been described in peer-reviewed work, even though the label "regression" has not. "Convention" means the age circulates widely but no source in our register publishes it as an event. Neither verdict tells you whether your baby is having a hard week. It tells you how much weight the chart deserves.

Durations are reported, not measured. Every figure in the third column reflects what parents describe rather than a tracked cohort, so treat each as a rough band with wide edges.

And nothing on this page diagnoses anything. A chart cannot distinguish a developmental shift from an ear infection. That distinction is a conversation with your pediatrician, and it is the one thing here you should not defer.

The chart

Scan the fourth column first, then read the row that matches your baby's age.

One row deserves flagging before you read it. Three to four months is the only entry on this table describing a permanent structural change rather than a passing phase. The acute disruption eases; the underlying architecture does not revert. Families waiting for sleep to go back to what it was at eight weeks are waiting for something that is not coming, and that wait is where most of the despair in the first year lives.

Three to four months: the only row that is architecture

Newborn sleep is not built like adult sleep, and around this age it is rebuilt.

In the early weeks infants alternate between active sleep and quiet sleep, precursors to REM and non-REM that do not yet show staged architecture. Newborns also enter sleep through active sleep, which is why a three-week-old can be fed to sleep, carried across a room and set down without noticing, and a five-month-old absolutely cannot.

Published work on early childhood sleep architecture describes active sleep becoming REM and quiet sleep becoming NREM after roughly two months, with distinct NREM stages emerging and sleep spindles expected on EEG by about three months. Sleep onset migrates as well: instead of dropping straight into an active state, the baby begins sleep in light NREM, and by around six months essentially all sleep begins that way. Cycles remain short by adult standards, so a baby surfaces toward waking repeatedly across a night. Underneath all of it the circadian system consolidates and a genuine day-night rhythm establishes.

The consequence at 2am is blunt. Your baby now notices. She surfaces at a cycle boundary, finds the room different from the one she fell asleep in, and asks for the original conditions back. Nothing broke — perception improved.

Which is why this row behaves differently from every other row on the chart. The acute stretch of frequent waking and short naps usually settles inside two to four weeks. The architecture never reverts. What helps is not a technique but a shift in what the falling-asleep conditions are: move them toward something that will still be true when she surfaces in the dark.

Six months: the weakest row on any chart

This entry survives mostly because charts prefer even spacing.

Six months is not associated with a distinct maturational shift the way three to four months is. The architectural reorganisation has already happened. The motor and cognitive changes that disrupt the second half of the first year mostly arrive later. Nothing in the register describes a six-month sleep event.

When sleep genuinely falls apart at six months, the cause is usually identifiable and specific: teething, a first illness as maternal antibodies wane, the introduction of solids, a house move, travel across time zones, or a schedule that quietly stopped fitting weeks earlier. That is good news, because a specific cause is something you can address, unlike a phase you can only outlast.

So if your six-month-old has fallen apart, resist the label and go looking. The most productive place to look is the day rather than the night — total daytime sleep, the timing of the last nap, and how long the final stretch before bed has grown. The 6 month old sleep schedule page lays out what the clock day should look like at this age, which makes drift much easier to spot than trying to reconstruct it from memory at 5pm.

Eight to ten months: motor rehearsal, and this part is on the record

This band produces the strangest night behaviour of the first year, and it is also the row where the mechanism is best studied after the four-month entry.

Babies rehearse new motor skills during sleep. A baby who learned to pull to stand on Tuesday will pull to stand in the crib at 1am on Wednesday, asleep in every sense that matters, and then wake genuinely distressed because she is upright with no idea how to get down. This is not a habit forming. Research has examined this directly: the emergence of pulling-to-stand has been studied in relation to sleep-wake regulation, and crawling has been examined specifically as a developmental transition mirrored in infant sleep. The clustering of disrupted nights around new gross-motor achievements is a real, investigated phenomenon rather than a parenting-forum theory.

The highest-yield response is entirely a daytime one. Drill the reverse of the skill on the floor — sitting down from standing, lowering from a pull-up — until it is automatic. The midnight version usually stops shortly after the daytime version becomes easy.

Running underneath is a cognitive change. Object permanence matures across the second half of the first year, and once it does, you leaving the room stops being neutral information. Separation protest at bedtime commonly sharpens here for exactly that reason.

The third element is logistical: daytime sleep need falls somewhere in this stretch, and a nap that is being fought or that pushes bedtime late has probably finished. Which nap and when is covered below.

Twelve months: a timing problem in a costume

The twelve-month row is where charts do the most damage, because the behaviour looks so much like readiness.

The pattern is familiar. Around a year, many babies start refusing the morning nap, or taking it and then refusing the afternoon one. It reads unmistakably as ready for one nap. For most twelve-month-olds it is not. It is far more often a timing mismatch: the morning awake stretch has grown, the first nap is now offered before enough pressure has built, and the baby is simply not tired enough to take it. Pushing that nap later often rescues the two-nap day and buys several more months of it.

Dropping a nap prematurely fails in a recognisable pattern — overtiredness by mid-afternoon, bedtime becomes a fight, night waking climbs, and early-morning waking appears. If those four arrived together shortly after you dropped a nap, the nap drop is your explanation, and it is reversible.

Walking and a jump in language comprehension are genuinely happening at this age and genuinely disturb sleep for a week or two. But they rarely account for a disruption running longer than that. If it has run three weeks, the chart is pointing you at the wrong thing and the schedule is where the answer is. The 2 to 1 nap transition page covers what an actual transition looks like as opposed to a bad fortnight.

Eighteen months: molars, autonomy, and a day that stopped fitting

By eighteen months the mechanism has shifted from biology to negotiation, which is why the responses that worked at four months are useless here.

This is the age a toddler discovers that bedtime is a place where she holds leverage. Stalling, requests, one more book, climbing out — all developmentally on time, none of it a sleep problem in the strict sense. It is a limits question, and it answers to an unhurried, entirely non-negotiable routine far better than to any adjustment of sleep timing.

Running alongside it, the genuine move from two naps to one lands for many children somewhere in this stretch, and it is legitimately hard because there is a period of weeks where one nap is not quite enough and two no longer fit. A small sleep debt runs while the day reorganises. There is no way to skip that middle, and expecting one is why this row gets reported as the longest of the lot.

Molars are the third strand and the one most often over-credited. They are real, and they are also permanently available as an explanation, which makes them a weak one. If disruption has run three weeks, something other than a tooth is driving it.

What the published sources actually support

This is the table that most charts leave out, and it is the reason this page exists. Claim by claim: is it published by a source we would cite, and if so, what does that source actually say?

The row worth dwelling on is the fourth. Percentages are the most common fabrication in this subject area — a figure for how many babies are affected at each age, or how many sleep through by a given month. No source in our register publishes any such number for these ages. Where you see one, it was invented or extrapolated, and its confident decimal point is doing rhetorical work rather than reporting a measurement.

Durations, and why every one of them is soft

The third column of the chart is the one people screenshot, and it is the one that deserves the least confidence.

None of those figures come from a tracked cohort. They come from what parents said afterwards, which is a genuinely useful signal and also a biased one: a fortnight that ended badly is remembered as longer than a fortnight that ended well, and a disruption that resolves in four days rarely gets reported at all. What survives into the charts skews long.

The underlying variation is enormous. A systematic review of normal sleep patterns in infants and children found wide spreads around every average at every age — which is the polite way of saying your baby may do none of this on schedule, and that this is unremarkable rather than a sign of anything.

So use the durations as a shape, not a countdown. "Two to six weeks" means most of these resolve inside a month and a half and some resolve in four days. It does not mean you should be measuring day nineteen against an expectation. The number that actually earns its place is the outer limit, and it gets its own section below.

Log five things before you trust any row

A chart can only tell you what typically happens at an age. What is happening in your house is a question the chart cannot answer, and four days of writing things down answers it better than any table.

The reason paper beats memory is that schedule causes are invisible from inside the day. At 5pm, running on five hours of sleep, nobody can reliably reconstruct whether the second nap has been creeping later all week. On paper it is obvious in thirty seconds. Log these five things for four consecutive days before you decide which row you are in.

Then read the log against the chart rather than the other way round. If a new skill appeared and the total sleep held steady, the developmental row probably fits. If no new skill appeared and the total has been sliding for a fortnight, the chart is not your answer — the day is.

• Morning wake time, to the minute, every day.
• Every nap start and end, including the failed attempts and how long they were fought.
• Bedtime, and how long the routine actually took from start to asleep.
• Every night waking with its clock time and what it took to resolve.
• Any new skill that appeared that week — rolling, sitting, crawling, pulling up, first words.

The six-week line

This is the most useful number on the page, and it is not in the chart.

A genuine developmental reorganisation ends. The capacity consolidates, the novelty wears off, and sleep reorganises around the new normal. If a disruption has run past about six weeks with no sign of movement, the regression framing has stopped being useful and is now actively in the way, because it tells you to keep waiting.

Past six weeks, treat it as one of two other things until proven otherwise. Either it is structural — the day drifted and never came back, a nap is now wrong, total sleep has slid outside the published range — in which case nothing will mature its way out of it and only a change will fix it. Or it is medical: reflux, an ear infection, a milk protein issue, iron deficiency, disordered breathing in sleep.

The second possibility is why this line matters more than any row above it. Persistent unexplained night waking, snoring, noisy or laboured breathing, pauses in breathing during sleep, or poor weight gain are conversations with your pediatrician rather than things to outlast. No sleep regression chart, this one included, can rule any of that out.

Where the chart runs out

Three honest limits, stated so the table above cannot be mistaken for more than it is.

The first is that a reference table flattens genuinely different events into one format. Four months is a permanent architectural change. Twelve months is usually a timing error. Eighteen months is partly a boundary question. Putting them in adjacent rows implies they are variations on one thing that all pass the same way, and they are not. The fourth column is there to interrupt that impression; the narrative version at baby sleep regression ages and stages argues it properly.

The second is that a great deal of what gets charted here is really schedule arithmetic. Total daily sleep, the timing of the last nap, and how long the final awake stretch has grown explain a large share of these weeks, and none of that is developmental. The homepage maps the schedule pages by age, and baby sleep schedule by age has the full chart of what the day should look like.

The third is that a baby who has none of these is not skipping a stage or storing it up. Population tendencies are not requirements. If your baby moved through four months, eight months and a year without a dramatic fortnight, nothing is wrong and nothing is pending.