Key Takeaways

“Sleep training” is an umbrella for four methods with different evidence: unmodified extinction, graduated extinction, bedtime fading, and parental presence. They are not interchangeable ¹.

A review of 52 studies for the American Academy of Sleep Medicine found that 94% reported behavioral methods worked, and more than 80% of treated children improved enough to count clinically ¹.

The harm question has been tested. In a randomized trial of 43 infants, cortisol fell slightly in both training groups compared with controls, and at 12 months there was no difference in attachment or behavior ².

The age that matters is around 6 months, and it comes from who was studied rather than a rule. A survey of 2,090 parents found real-world use starting at 5.3 months on average, which its authors called earlier than generally recommended ³.

Two official sources sound like opposites. The AAP’s parent page calls the idea that self-soothing means cry it out “absolutely not true” ⁴; one NHS trust says it does not recommend leaving a baby to cry ⁵. They are describing different things.

The short answer

Sleep training is the practice of changing what a baby needs in order to fall asleep. A baby who is fed or rocked to sleep learns that arrangement as the entry condition for sleep. Training changes the condition, so the baby can do the last part alone.

That definition matters because it separates two things parents often merge. Training is not night weaning. A 7-month-old can learn to settle without a feed at bedtime and still take two feeds overnight. None of the trials below required anyone to drop them.

It is also not a newborn project. The randomized evidence starts at 6 months, and the largest real-world survey put the average start at 5.3 months ²³. If your baby is younger than that, the sections on the age floor and on what to rule out first are the ones that apply to you.

Four methods, not one

The search results treat “sleep training” as a single decision, usually with a chart. It is four decisions, and the evidence is not evenly distributed across them.

Method	What you actually do	Evidence
Unmodified extinction	Put the baby down awake and do not go back in until a set time or a feed	Strongest in the review, supported by controlled group studies ¹
Graduated extinction	Put the baby down awake, then check at increasing intervals with brief, boring comfort	Supported, and tested in a randomized trial ¹²
Bedtime fading	Move bedtime later for a few nights so sleep pressure is high, then shift it back earlier	Supported, and tested in the same trial ¹²
Parental presence	Stay in the room and offer less and less help over successive nights	Not among the treatments that review found support for; surveyed in the real-world study ¹³

The fourth row is the one most often marketed as the gentle option. It is also the row with the thinnest trial base. That is not an argument against it. It is a reason to be suspicious of any page that presents it as the safe option and the others as the harsh ones, because the ranking is invented.

The “Ferber chart” that fills the top search results is a schedule of checking intervals. The trials tested graduated extinction as a package, not one interval schedule against another ². A chart from a book is a starting point, not a finding.

What the trials measured

Three studies carry most of the weight here, and each answers a different question. One reviews what has been tried. One tests what happens inside a family. One surveys what families actually do.

Study	Design	What it found
Mindell et al., 2006	Review of 52 treatment studies for the American Academy of Sleep Medicine	94% reported the behavioral methods worked; over 80% of treated children improved clinically, maintained for 3 to 6 months ¹
Gradisar et al., 2016	Randomized trial, 43 infants aged 6 to 16 months, three arms	Large drops in sleep latency in both training arms; large drops in night wakings and time awake after sleep onset in the graduated extinction arm ²
Kahn, Barnett and Gradisar, 2023	Survey of 2,090 US parents, infants 3 to 18 months	64% had used a method; extinction rated harder but more helpful and faster; night sleep longer and more consolidated ³

The 2006 review is the reason the two extinction methods sit at the top of the earlier table. Its task force found strong support for unmodified extinction and for preventive parent education, and support for graduated extinction, bedtime fading with positive routines, and scheduled awakenings ¹.

The 2016 trial is the one that measured a biological marker rather than asking parents how they felt. Infants gave saliva samples in the morning and afternoon, which captures the resting pattern rather than the peak during an episode. Cortisol showed small to moderate declines in the graduated extinction and bedtime fading groups compared with controls ². Mothers’ stress fell a little over the first month, with no measurable change in mood. That direction matters, because the common worry is that training raises stress, and the measurement went the other way.

The 2023 survey is the reality check. Extinction methods were rated significantly more difficult to carry out than parental presence, and also more helpful, shorter, and quicker to show improvement ³. That is a coherent trade rather than a contradiction. The method parents rate as hardest is also the one they rate as working fastest.

Two limits apply to all three. The randomized trial had 43 infants, which is enough to detect a large sleep effect and not enough to rule out small ones. The survey used parent questionnaires plus exploratory camera-based sleep estimates, so it describes associations rather than causes ³. Neither was designed to answer whether training is right for your family.

Does it harm attachment?

This is the question that decides the issue for most parents, and it has better evidence than the rest of the topic. Three separate designs point the same way.

The clearest test comes from a UK study of 178 infants followed to 18 months. Leaving a baby to cry occasionally or often during the first 6 months was not associated with any adverse effect on attachment or on behavioral development ⁶. Attachment was measured with the Strange Situation procedure, the standard lab test for how a toddler uses a parent for reassurance. The same study found that leaving a baby to cry at term was linked to less frequent crying at 3 months. Doing so a few times at term, and often at 3 months, was also linked to shorter fuss and cry duration at 18 months. For what the crying curve looks like with no intervention at all, the guide to normal crying has the numbers.

The randomized trial adds a second check. Twelve months after the intervention, mothers completed assessments of emotional and behavioral problems, and the mother and child pairs went through the Strange Situation. No significant differences appeared in either ². The authors’ conclusion is unusually plain: both methods “provide significant sleep benefits above control, yet convey no adverse stress responses or long-term effects on parent-child attachment or child emotions and behavior” ².

The survey adds a third, weaker check. It found no differences between the method groups in parent sleep, daytime sleepiness, depression, or bonding with the baby ³.

Now the limit. The trials above stop at 12 months of follow-up ². One went further: a 5-year follow-up of 326 children whose parents were taught behavioral techniques at 8 to 10 months found no differences at age 6 in emotional or conduct scores, sleep problems, chronic stress, the child-parent relationship, or maternal depression and anxiety ⁷. Its authors concluded that the techniques have “no marked long-lasting effects (positive or negative)” ⁷.

That is the longest randomized check anyone has published, and it reaches age 6. “No detectable harm in the studies that exist” remains the accurate statement. It is not the same as “harmless,” and no study has shown the absence of an effect it was not designed to detect.

One wrinkle matters here. The UK study found no adverse effect from leaving a baby to cry during the first 6 months ⁶. That runs against the usual advice to wait until 6 months. It is one observational study, and it means the age floor below is a convention rather than a bright line.

The age floor

The floor is real, and it is softer than the internet makes it sound.

What is solid: the randomized evidence begins at 6 months ². The task force review covered treatment studies from birth through age 4 ¹. That range is wider than any single trial’s, so it does not settle a minimum age on its own. And the authors of the 2,090-parent survey described real-world use at an average of 5.3 months as “pervasive and occurs earlier than generally recommended” ³. That phrasing implies a general recommendation exists, and that it sits later than 5.3 months.

What is not solid: a specific night. There is no study showing that training works at 6 months and one week and fails at 5 months and three weeks. The reason is practical. Under 4 months, night waking is usually a feed or a comfort need, which is why the AAP’s parent guidance describes a response ladder rather than a training plan ⁴.

The practical version is this. Before 4 months, treat night waking as communication and use a soothing ladder rather than a training plan. That is also what the AAP’s parent guidance sets out ⁴. The sequence runs from eye contact, to voice, to a hand on the chest, to holding, rocking, and feeding. The goal is to calm the baby with less help each time. Between 4 and 6 months, the deciding factor is usually whether the baby still needs the night feeds. From 6 months, the evidence above applies.

The bands look like this in practice.

Age	What night waking usually means	What to do
Under 4 months	Feeds, and a need for help settling	Use the soothing ladder, not a plan ⁴
4 to 6 months	Feeds if the baby still wakes hungry; a habit starting to form	Decide whether night feeds are still needed
6 months and over	Often habit as much as hunger	The trial evidence above applies ¹²
Any age, during illness or a regression	Symptoms, or a developmental change	Wait until it passes

One timing warning. It is usually worth waiting out the 4-month sleep regression before starting. The disruption is developmental and tends to settle without help, which makes it a poor test of any method. If you are in the middle of it, the 4-month sleep regression guide is the better page, and the wake windows guide explains the daytime side of the same problem.

Why the advice depends on who you ask

Two authoritative sources say things that sound like opposites, and the reason is that they are answering different questions.

The AAP’s page for parents is about the word “self-soothing.” It says the term “can sometimes be interpreted as a parent making a baby ‘cry it out’ or ignoring their cries. This is absolutely not true” ⁴. Read closely, that is a claim about a technique: teaching a baby to settle is not the same as ignoring a baby. It is not a claim that leaving a baby to cry never helps.

An NHS trust leaflet for parents of 6 to 12 month olds goes further in the other direction. “As a service we do not recommend any sleep modification/training techniques that involve leaving your baby/child to cry, such as controlled crying, pick up/put down” ⁵.

Both statements are defensible and they are not actually in conflict. The AAP is describing a skill and a ladder of responses. The trust is describing a practice and choosing not to endorse it. One is about what self-soothing means, the other about what a service will recommend.

The practical consequence is easy to miss. If your health visitor follows a policy like that one, you may be steered away from any method that involves crying. The trial evidence says those methods work and show no measured harm at 12 to 18 months ²⁶. Neither position is careless. The leaflet states a service position without saying which evidence it weighed, and careful people hold both views.

If the two disagree, ask each what they are basing it on. Usually one is citing trials and the other is citing a service policy, and that tells you what you are actually choosing between.

What to rule out, and what to skip

Before any method, check whether the waking has a cause that training cannot fix. A plan aimed at the wrong problem fails, and then the method gets blamed.

What it looks like	What it may be	What to do instead
Waking every 2 to 3 hours, feeding eagerly	Hunger, which is normal at this age	Keep the feeds, train only the bedtime settling
Waking with arching, hiccups, or sour breath	Reflux	Treat the reflux first
Sudden waking with a new cry, hands at the mouth	Teething or an ear infection	Pain first, plan later
Waking more after a short-nap day	Accumulated sleep debt	Fix the day before the night

The sleep schedule by age guide covers the daytime side of that last row, and it is the first thing to check if the nights got worse without any other change.

Then the things to skip, which are less about method and more about safety.

Weighted sleep sacks and weighted swaddles are the clearest one. A 2024 statement from a commissioner at the Consumer Product Safety Commission quotes the agency’s warning, “Don’t use weighted blankets or weighted swaddles.” The same statement records that the AAP, the NIH, and the CDC have all warned against weighted infant products, and that the commissioner is aware of multiple infant deaths involving weighted sleep sacks ⁸.

Cereal in the bottle, pillows, wedges, and positioners in the crib all fail for the same reason. They add a hazard for a benefit that has not been demonstrated. The 2022 safe-sleep guidance is that the sleep surface should be firm, flat, non-inclined, and “clear of everything but a fitted sheet” ⁹. The flat head guide covers why the pillow version of this idea is a bad trade in more detail.

Also skip: training a baby who is unwell, training during travel, and any plan that involves placing a baby on the stomach to sleep. None of those are methods. They are ways to make a bad week worse.

Setting up the first night

The method is a small part of this. Most of what determines whether the first week works is decided before bedtime.

Fix the environment. Dark, cool, and boring. Firm flat surface, fitted sheet, nothing else. If the room is too warm, the baby wakes more. Log the nights as you go, because memory is unreliable after a week of broken sleep and the pattern is the evidence that the plan is working. The sleep tracker keeps night wakings and feeds in one place.

Build a short routine. Three or four steps, same order, about 20 minutes, ending in the room where the baby sleeps. The routine is a signal, not a ritual, and a long one gives a tired baby more chances to become overtired.

Put the baby down drowsy but awake. This is the actual mechanism. A baby who is already asleep when placed has nothing to practice. A baby who is calm but awake has the whole task in front of them, which is the point.

Choose one method and write down the response plan. Which method, what you will do at 10 pm, what counts as a feed, and who is on duty. Two parents running different plans on alternate nights is the most common reason a first attempt fails. The caregiver handoff guide covers how to keep the plan identical when someone else takes the night.

Decide your stop rule before you start. Fever, vomiting, an unusual cry, or a baby who will not settle at all are reasons to stop and check in, not reasons to push through. Writing that down in advance is easier than deciding it at 3 am.

When it isn’t working

Give it two weeks. The trials measured improvement within weeks, and parents rated extinction methods as quicker to show results than parental presence ²³. If nothing has moved in two weeks, the problem is usually the plan, the timing, or the diagnosis, in that order.

Stopping is a legitimate outcome. A method that is making a parent worse is not working, even if the baby’s sleep numbers improved, and no study has shown a benefit that justifies a household in crisis.

Get advice rather than another method if the baby is not gaining weight, has fewer wet diapers, screams in a way that does not stop, or has symptoms that suggest reflux or breathing trouble. Those need a clinician, not a schedule. And if you are the one who is not coping, say so at the next visit. Persistent sleep deprivation is a reason to ask for help on its own, separate from anything about the baby.

Frequently asked questions

What age can you start sleep training?

The randomized evidence begins at 6 months ², and researchers in the field describe earlier use as earlier than generally recommended ³. Before 4 months, use a soothing ladder instead. Between 4 and 6 months it depends on whether the baby still needs night feeds.

Is it safe to let a baby cry?

For infants over 6 months, the studies that measured it found no adverse effect on attachment or behavior at 12 to 18 months, and cortisol fell rather than rose in the randomized trial ²⁶. That is reassuring rather than conclusive. The longest randomized follow-up reached age 6 and found no differences ⁷.

Does sleep training mean night weaning?

No. Training changes what a baby needs to fall asleep at bedtime. Dropping night feeds is a separate decision, and the trials did not require it ²³. Many families keep one or two night feeds throughout.

Does the Ferber method work?

Graduated extinction has support in the 2006 review and was one of the two arms in the 2016 trial, where it produced large drops in night wakings ¹². The specific interval chart is not what was tested. The trials tested the method as a package.

Disclaimer: This article is for informational purposes only and does not constitute medical advice. Sleep and feeding needs vary between babies, and the studies here describe groups rather than your child. Talk to your pediatrician or health visitor before starting any sleep plan, and particularly if your baby was premature, is not gaining weight, or has reflux or breathing problems. Seek urgent care for a baby who is hard to wake, feeding poorly, or has a fever.

Mindell JA, Kuhn B, Lewin DS, Meltzer LJ, Sadeh A; American Academy of Sleep Medicine, “Behavioral treatment of bedtime problems and night wakings in infants and young children,” Sleep, October 2006. Source ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹
Gradisar M, Jackson K, Spurrier NJ, Gibson J, Whitham J, Williams AS, Dolby R, Kennaway DJ, “Behavioral Interventions for Infant Sleep Problems: A Randomized Controlled Trial,” Pediatrics, June 2016. Source ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸
Kahn M, Barnett N, Gradisar M, “Implementation of Behavioral Interventions for Infant Sleep Problems in Real-World Settings,” Journal of Pediatrics, April 2023. Source ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹
Ideran P, Fishbein M, “Self-Soothing: Help Your Baby Learn This Life Skill,” HealthyChildren.org (American Academy of Pediatrics), accessed October 2026. Source ↩ ↩² ↩³ ↩⁴ ↩⁵
Wirral Community Health and Care NHS Foundation Trust, “Your baby’s sleep at 6 to 12 months.” Source ↩ ↩²
Bilgin A, Wolke D, “Parental use of ‘cry it out’ in infants: no adverse effects on attachment and behavioural development at 18 months,” Journal of Child Psychology and Psychiatry, November 2020. Source ↩ ↩² ↩³ ↩⁴
Price AMH, Wake M, Ukoumunne OC, Hiscock H, “Five-year follow-up of harms and benefits of behavioral infant sleep intervention: randomized trial,” Pediatrics, October 2012. Source ↩ ↩² ↩³
US Consumer Product Safety Commission, Commissioner statement on weighted infant products, April 2024. Source ↩
American Academy of Pediatrics, “Sleep-Related Infant Deaths: Updated 2022 Recommendations for Reducing Infant Deaths in the Sleep Environment,” Pediatrics, July 2022. Source ↩