Cry It Out: What the Evidence Says

Cry it out, in the literature, means unmodified extinction: the routine, then down awake, then no return until morning except to check that she is safe. It has the firmest efficacy support of any method in the strongest document available, and the least palatable protocol of the six. Both of those are true at once, and this page is written to hold them together rather than to pick one. It is one method among several, and the methods compared page sets out the rest.

This site takes no position on whether to use it. What it can do is separate what was measured from what is asserted, in both directions, because almost everything written about this method is written by somebody who already decided.

Most people saying it are not doing it

The first correction is a definitional one, and it changes what half the arguments are about. Unmodified extinction means not going back in. Graduated extinction means going back in on a rising schedule of intervals — that is what the Ferber method is, and it is what most families mean when they say they are doing cry it out.

The two were tested separately, and only one of them means genuinely leaving her until morning. So when somebody tells you what a study found about cry it out, the first question is which protocol the study ran. The trial that measured cortisol and attachment, quoted by both sides of this argument constantly, was a trial of graduated extinction and bedtime fading. It is not a study of unmodified extinction at all.

What the efficacy evidence says

The task force appointed by the American Academy of Sleep Medicine went through 52 treatment studies of bedtime problems and night wakings in infants and young children. It reported that 94% of them found the behavioural interventions efficacious, that over 80% of treated children showed clinically significant improvement maintained for 3 to 6 months, and that the strongest support went to unmodified extinction and to preventive parent education, with further support for graduated extinction, bedtime fading with positive routines and scheduled awakenings.

Read the numbers precisely. Over 80% is most children, not all of them. Clinically significant improvement is not the same as a baby who no longer wakes. Three to six months is the follow-up the review reports, not a claim about a childhood. And it is a statement about groups of children rather than a prediction about one — yours is not in it.

Set beside that the trial with the most honest arc in this subject: 156 mothers of infants aged 6 to 12 months with severe sleep problems by parental report, given a behavioural intervention over three consultations or written information about normal sleep. At two months the intervention had resolved more sleep problems, 53 of 76 against 36 of 76. By four months the difference had gone. What did hold was the mothers' own mood: depression scores fell further in the intervention group, and for mothers scoring 10 or more on the Edinburgh postnatal depression scale that difference was still there at four months.

Those two results sit together comfortably once you notice what each is measuring. Methods work, on average, for a while; the untreated families catch up; and the parent-facing benefit is the one that lasted longest in the trial that looked for it.

The harm argument, examined rather than answered

The sentence at the centre of this is that sleep training damages your baby, usually followed by an account of cortisol and toxic stress. It deserves an examination rather than a slogan, in either direction.

Here is what was actually measured. One randomised trial enrolled infants aged 6 to 16 months, allocated them to graduated extinction, bedtime fading or a sleep-education control, and took salivary cortisol through the intervention. Cortisol declined modestly in both active groups — down, not up. At the twelve-month follow-up there were no differences in parent-child attachment measured by the strange situation, and no differences in the children's emotional and behavioural problems.

A second trial went further out. Children whose sleep problems at seven months had been randomised to a behavioural intervention or to usual care were compared at age six on mental health, sleep, psychosocial functioning, stress regulation, the child-parent relationship and maternal mental health. There was no evidence of differences on any of those outcomes, and the authors concluded that behavioural sleep techniques have no marked long-lasting effects, positive or negative.

Now the limits, because they are the honest half. The samples were modest. The follow-up periods were finite — a year in one, to age six in the other — and a childhood is longer than either. Nobody has measured every outcome anybody might care about. And absence of a found effect is weaker than proof of no effect, which is the sentence this whole section turns on, because it cuts both ways: it means the harm claim is unsupported, and it means the reassurance cannot be absolute either. Anyone offering you certainty in either direction is offering you something the evidence does not contain.

One more limit worth naming: none of the trials above ran unmodified extinction on its own with cortisol attached. The measurement everybody cites was taken during graduated extinction. Applying it to a stricter protocol is an inference, not a finding.

What it asks of you, and who it does not suit

The protocol is the least palatable part and there is no way to write it kindly. You do the routine, put her down awake, and do not return — you check she is safe, and that is all the night contains. Whatever you have decided about night feeds happens on a plan set before bedtime rather than in response to crying, and that decision is a matter for you and your pediatrician or health visitor, since nothing on this site attaches a month to night feeds.

It does not suit a household that cannot hold the same rule for a fortnight, and that is the most common mismatch by far: half a method, run inconsistently, teaches nothing except that persistence pays. It does not suit two parents who disagree about it, because one of them will go in. It does not suit the week of an illness, a house move or a holiday. And it does not suit a family who will find the nights unbearable — that is a legitimate reason, needs no further justification, and the gentler options have evidence of their own, including bedtime fading, which was one of two active arms in a randomised trial.

No trial publishes a reliable figure for how long the crying lasts, and the numbers you have read — forty-five minutes the first night, less the second — are convention rather than measurement. Nor should any trial be read as a description of everybody who started it: families do stop partway through in this literature, and none of the sources on this page publishes a dropout figure, so treat any you are quoted as unsourced. Decide your own limit in the evening, write it on a printed night sheet beside the times, and treat reaching it as information rather than failure. If you want intervals instead, the check-in chart ships its numbers as editable defaults, because no publisher this site is allowed to cite reprints them.

The thing neither side says

Waking in the night is ordinary. The AAP's own description of a good sleeper at this age is a child who wakes frequently but can get herself back to sleep — not a child who sleeps ten unbroken hours — and it calls frequent waking developmentally appropriate. Nothing on this page implies that a baby who wakes has a problem, or that a family who chooses no method at all has failed to do something.

What the argument usually leaves out is that the decision is not only about the baby. The one outcome that held longest in the trial above was the mothers' mood, and a household running on four hours of sleep is a fact about a child's life too. That is not an argument for the method. It is an argument for counting the whole household when you weigh it, which is a judgement rather than a finding, and yours to make.

What does not change, and when to stop

However the night is run, the sleep itself starts the same way. 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, with nothing weighted on or over them and nothing in the cot that could cover her face. The safe sleep basics page has the full version, and it outranks every method page on this site including this one.

Stop for the night if she seems unwell rather than cross: a temperature, a cry that does not sound like her usual one, a nappy, a leg caught in the bars, anything that hurts. A baby who cannot be settled at all, who is unusually sleepy, feeding poorly or not gaining weight is not a method question, and neither is a worry that will not leave you. Those are conversations with your pediatrician or health visitor. This page is general information, not medical advice, and it cannot see your baby.