When to Start Sleep Training

Every randomised trial in this subject enrolled babies of six months or older, so six months is where the evidence starts — not because anybody drew a line there, but because that is where the researchers looked. One trial randomised infants aged 6 to 16 months; another enrolled infants of 6 to 12 months whose parents reported severe sleep problems; the trial with the five-year follow-up recruited at 7 months. Below six months, nothing has been published either way. That is not a prohibition and it is not permission. It is a gap, and this page would rather show you the gap than fill it with confidence.

Almost everything else written about the age question is somebody's opinion dressed as a threshold — four months, sixteen weeks adjusted, thirteen pounds. None of those numbers has a publisher behind it. What follows is what the trials actually did, what the AAP does say about timing, and how to think about readiness when there is no checklist to consult.

The ages on the trial labels

Three studies carry this subject, and the ages they enrolled are the most useful thing about them.

The first randomised infants aged 6 to 16 months to graduated extinction, bedtime fading or a sleep-education control, measured salivary cortisol through it, and followed the children for a year. The second enrolled 156 mothers of infants aged 6 to 12 months with severe sleep problems by parental report, and compared a behavioural intervention discussed over three consultations against written information about normal sleep. The third identified sleep problems at 7 months, randomised the families, and went back at age six to look for lasting harm and lasting benefit.

Six, six and seven months. That consistency is worth noticing, because it means every finding you will read about whether these methods work, and every finding about whether they do damage, was produced in babies of at least half a year. The evidence does not stop at sixteen months — the task force review of 52 treatment studies covers infants and young children rather than babies alone — but it genuinely does start at six.

An enrolment age is not a safety threshold

This is the distinction the whole page turns on. Researchers chose six months for practical reasons: younger babies feed more at night, their sleep is still organising, and a trial that enrolled them would be measuring feeding as much as settling. That choice tells you where the knowledge is. It does not tell you that something changes on a baby's six-month birthday, and it does not tell you that a five-month-old would be harmed.

So there are two honest sentences and this site will only say those two. Above six months, there is evidence, and the comparison of the methods lays out what each piece of it covers. Below six months, nobody has looked, and anybody who tells you otherwise is quoting a consultant rather than a trial.

Under four months there is nothing at all

Younger again, and the picture changes from a gap in the evidence to a difference in the baby.

Babies do not have regular sleep cycles until about four months, according to the AAP, and although newborns sleep about 16 to 17 hours a day they may only sleep one or two hours at a time. There is no consolidated night to work on. Feeding drives the first weeks, waking is how a small baby eats, and a method that assumes settling is a learned pattern has nothing to attach itself to yet.

The AAP's only remark about timing in this area is a small one: from about four months you can try putting a baby to bed drowsy rather than asleep. That is a suggestion about how you put her down, not a programme, and plenty of babies have not read it. For the months before that, the honest tools are a rhythm rather than a clock, daylight in the day and dullness at night, and as much help as you can get hold of. Day and night confusion covers the first of those and the bedtime routine covers the second — and the routine is the one intervention here with a randomised trial to itself, no crying in it, and nothing to decide.

What "ready" honestly means

There is no published readiness checklist. Every version you find — including the sensible-sounding ones — is somebody's opinion, and the honest way to read them is as a description of the households where these attempts tend to go well rather than as a test a baby passes.

What families usually mean by ready is a baby past four months whose sleep has organised into cycles, who is feeding well and gaining weight, who is well this week; and a household that can do the same thing every night for a fortnight. Notice that only the first half of that is about the baby. The second half is about you, and it is the half that decides most attempts.

The one thing worth doing before anything else is looking at the day. A bedtime that lands two hours after the last nap, or five hours after it, will produce a bad evening whatever method is or is not attached to it — the wake window calculator gives you the band the last stretch before bed should sit in, and wake windows by age explains where those bands come from and why they are bands. Fixing the timing sometimes removes the problem you were about to train.

The weeks not to start in

Illness. The week teeth are visibly coming through. A house move, a holiday, the fortnight after a new sibling arrives, the week somebody goes back to work. And any stretch where a skill is plainly arriving — rolling, crawling, pulling up, first words.

The reason is not that any of those makes an attempt harmful. It is that during any of them you cannot tell whether a change worked, and telling whether it worked is the entire point of making one. A fortnight with three variables in it produces no information, just a tired household and a theory. The timeline of the harder stretches marks the ages where nights commonly come apart on their own, and a printed day sheet filled in for a week beforehand will show you whether the week ahead is calm enough to read.

If a fortnight like that never seems to arrive, that is ordinary rather than a failure. Clear fortnights are rarer than the advice assumes, and waiting for a perfect one is its own kind of trap.

Not choosing is a complete answer

Nothing on this page argues that a family should sleep train, and nothing argues that they should not. It is worth saying plainly, because the two loudest positions in this subject both treat the question as settled and it is not.

Night waking is developmentally ordinary. The AAP's own description of a good sleeper in the first year is a child who wakes and gets herself back to sleep, and it is normal for a six-month-old to wake in the night and settle again after a few minutes. A household that decides to ride it out is not storing up a problem, and a household that decides to change something is not doing damage. The evidence is here so a decision can be informed, not so one can be sold.

What does not change either way is the sleep itself: babies go down on their back for every sleep, day and night, on a firm flat mattress, in a cot clear of blankets, pillows, bumpers and soft toys, in your room for at least the first six months. Nothing weighted goes on or over a sleeping baby, at any age and during any method. Safe sleep basics is the page that outranks all of this.

When the age is not the question

Some nights are not a timing problem. A baby who cannot be settled at all, who is feeding poorly or not gaining weight, who is unusually sleepy or hard to rouse, whose cry does not sound like her usual one, or who seems to struggle for breath in her sleep needs a person rather than a plan. So does a parent who is not coping, and that call is worth making today rather than after another fortnight of trying harder. Speak to your health visitor or pediatrician. This page is general information, not medical advice, and it cannot see your baby.