How to Sleep Train a Baby — Every Method, Honestly Compared

This is a choice, not a milestone

Sleep training is not a developmental stage your baby has to pass through, and skipping it does not leave a gap. Large parts of the world do not do it at all. Cross-cultural work on infant sleep consistently finds substantial variation in normal practice — in some settings solitary infant sleep and independent settling are the expectation, in others close contact through the night is, and children in both grow up fine.

This matters because the framing you encounter online often implies otherwise. "Sleep training" gets discussed like tummy time or solids: a thing you do at the right age or fall behind. It is not that. It is one option for addressing a specific problem — a family that is not getting enough sleep — and it competes with other options, including changing the schedule, changing who does the nights, changing sleeping arrangements, or waiting.

So the honest first step is not choosing a method. It is deciding whether you have a problem worth intervening in. Some families are functioning fine with a baby who wakes twice a night and would rather not change anything. Some are three months into severe deprivation and need something to change urgently. Those are different situations and they warrant different answers.

If you are here because someone told you that you should be doing this by four months, or six, or that you have missed a window: there is no window. Behavioural sleep interventions have been studied and used successfully across infancy and toddlerhood. Nothing closes.

What sleep training actually means

The term covers a wider range than most people assume, and a lot of arguments about it are really arguments about definitions.

At its narrowest, sleep training means a structured behavioural intervention designed to help a baby fall asleep at bedtime, and return to sleep after a night waking, without an adult reinstating the conditions they fell asleep in. That is the technical target: not the elimination of waking, which is neither possible nor desirable, but the reduction of signalling — the calling-out that follows a normal end-of-cycle arousal.

At its broadest, the term is used for anything that improves sleep on purpose: establishing a consistent bedtime routine, darkening the room, moving to an age-appropriate schedule, introducing a consistent sleep environment. Some of these are recommended by essentially everyone, including people who oppose sleep training as narrowly defined.

That distinction is useful because the broad interventions carry very little controversy and often do most of the work. A consistent, unhurried bedtime routine performed in the same order every night has good evidence behind it and asks nothing of anyone that feels difficult. A great many families who think they need a formal method actually have a schedule problem, and fixing the schedule resolves it without any crying protocol at all.

Which is the genuinely useful sequencing advice on this page: exhaust the boring interventions first. They are free, uncontroversial, and frequently sufficient.

What has to be true first

No method works against a baby whose underlying needs are not being met, and attempting one in that situation produces a lot of crying and no result, which is the worst of both worlds.

Age is the first consideration. Most published protocols were studied in infants of around six months and older; the evidence base below six months is much thinner, and below four months there is effectively no case for it — sleep architecture has not matured, the circadian rhythm may not be established, and night feeds are typically necessary. Guidance generally advises waiting until at least four months, and confirming readiness with your pediatrician rather than an age chart.

Feeding is the second, and it is the one most often got wrong. Many babies genuinely need one or more night feeds well past six months, and there is no universal age at which they stop. Whether your baby still needs to eat overnight is a question about their growth and intake, which is a pediatrician question. Plenty of families sleep train while keeping one or two scheduled night feeds, and that is a perfectly coherent plan.

Health is the third. Reflux, ear infections, eczema, food-protein intolerance and sleep-disordered breathing all present as unexplained night waking, and none of them respond to behavioural intervention. If waking is accompanied by snoring, mouth-breathing, pauses in breathing, pain signals, or poor weight gain, that is a medical evaluation, not a training problem.

And fourth, timing. Starting during a genuine developmental disruption, an illness, a house move or a trip loads the dice against you. Most guidance suggests waiting for a reasonably ordinary fortnight.

• Baby is at least 4 months, and ideally closer to 6, with pediatrician sign-off
• Night feeding needs have been discussed with your pediatrician and a plan agreed
• No current illness, teething flare, ear infection, or reflux under investigation
• Not mid-move, mid-travel, mid-time-zone-change, or in the acute phase of a developmental leap
• Schedule and wake windows are already age-appropriate — check the baby sleep schedule by age page
• Both caregivers are genuinely on board; inconsistency between adults is the most common reason methods fail
• You have a clear window of 1–2 weeks where you can stay consistent

Graduated extinction (the Ferber approach)

The most widely studied and most widely used method. The baby is put down awake, the parent leaves, and returns at progressively lengthening intervals to briefly reassure without picking up or feeding.

A typical first night uses intervals of roughly three, five and ten minutes, extending on subsequent nights. Checks are short — a minute or less — and deliberately dull: a hand on the back, a few words, and out. The point of the check is not to stop the crying. It is to confirm to both parties that nobody has been abandoned. Parents who expect the check to soothe the baby into sleep usually find it appears to make things worse and abandon the method on night two.

The intervals are not the active ingredient and there is nothing magic about the specific numbers. What matters is that the response is predictable and does not escalate. Many families use fixed intervals throughout rather than lengthening them, and the research does not suggest the lengthening is essential.

Typical timeline: most families who see it work see meaningful change within three to seven nights, often with a marked worsening on night two or three before improvement. The night-two spike is well known enough to be worth planning for, because it is the point at which most people quit.

What it asks of you: tolerance for crying that you are choosing not to fully resolve, for several nights. That is a real ask and it is not equally tolerable for everyone. Deciding it is not for you is a legitimate conclusion, not a lack of resolve.

Full extinction

Sometimes called unmodified extinction, and in popular usage "cry it out" — although that phrase is used so loosely that it has stopped being informative. Here it means the specific protocol: baby is put down awake, and the parent does not return until morning, except to respond to genuine need or distress signals that indicate something is wrong.

In the research literature this is among the best-supported approaches in terms of measured effectiveness, and it tends to produce results faster than graduated methods — often in fewer nights. It is also the one most parents find hardest to carry out, and abandonment partway through is common. That is worth stating clearly, because a method abandoned on night three is generally worse than a method never started: the baby has experienced the change without the resolution, and the parent has confirmation that it "does not work."

Most families who use it modify it — keeping a monitor on, agreeing a threshold at which they will go in, maintaining a scheduled night feed. Those modifications do not ruin it. There is no purity requirement here.

The practical distinction between this and graduated extinction is smaller than the emotional distinction. Both involve a period of crying that the parent chooses not to end immediately. If that period is unacceptable to you, both are unacceptable, and the useful next step is to look at the gradual approaches below rather than to look for a gentler version of the same thing.

The chair method

Also called camping out or the sleep lady shuffle in various formulations. The parent stays in the room, sitting in a chair beside the crib, and offers minimal verbal reassurance without picking the baby up. Every few nights the chair moves further toward the door, until after roughly one to two weeks it is outside the room.

The appeal is obvious: the parent is present throughout, and the change happens by degrees rather than all at once. For families who find the idea of leaving intolerable but want the baby settling independently, it is often the compromise that makes the project possible at all.

The trade-offs are real too. It takes considerably longer — commonly two to three weeks against three to seven nights — and it requires a lot of sustained parental presence at a time of day when you are least equipped for it. Some babies also find a visible, non-responsive parent more provoking than an absent one, and cry more with the chair method than they would with graduated checks. If you try it and the crying is escalating rather than settling by night four or five, that may be what is happening.

A typical progression: nights one to three beside the crib, four to six at the midpoint of the room, seven to nine by the door, ten to twelve in the doorway, then out. Move at whatever pace is holding, and do not move back to a closer position after a bad night — that turns the progression into a lottery.

Pick-up-put-down

The most hands-on of the structured approaches. The baby is put down awake; when they become upset, they are picked up and comforted until calm, then put back down while still awake. Repeat, potentially many times.

It is popular because it involves the least crying-without-response of any of the methods, and for parents whose objection is specifically to unresolved distress it is often the only one that feels acceptable.

It is also, in practice, the most demanding to execute and the least reliably effective. Sessions of forty-five minutes and dozens of repetitions are common. It tends to work best with younger babies, roughly four to seven months; with older, heavier, more determined infants the repeated pick-ups can become stimulating rather than settling, and some babies escalate specifically because being picked up is now on offer.

If you are going to use it, two things help. Put the baby down at the point of calm rather than the point of sleep — putting down an asleep baby simply restarts the cycle at the next arousal. And set an advance limit on the session, both for your own sake and because a two-hour battle is not producing learning for anyone.

It is worth knowing that this approach has less published evidence behind it than graduated extinction or fading. That is not evidence it does not work; it is a genuine gap. Many families report success with it. The research simply has not concentrated there.

Bedtime fading

The least discussed method and, for a specific problem, often the most effective one — which is why it deserves more attention than it usually gets.

Bedtime fading targets a particular situation: a baby who fights bedtime at length, taking forty-five minutes or an hour to fall asleep, and who is genuinely not tired at the time they are being put down. The intervention is to move bedtime temporarily later — to the time the baby is actually falling asleep — so that they go down when sleep pressure is high, fall asleep quickly, and the association between the crib and prolonged struggle is broken. Once falling asleep is happening within about fifteen minutes reliably, bedtime is moved earlier in small increments, often fifteen minutes every few nights, until it reaches the target.

What makes it notable is that it was one of two interventions tested in the randomised trial by Gradisar and colleagues published in Pediatrics in 2016, alongside graduated extinction, and both were found to reduce time to fall asleep relative to control. It performs, and it involves markedly less crying than the extinction-based approaches — often almost none, because the baby is genuinely sleepy when put down.

It does not directly address night waking, which is its main limitation. And it requires an accurate read on when your baby is actually falling asleep, which means keeping a log for several nights first.

For families whose primary complaint is bedtime resistance rather than night waking, this is frequently the right first thing to try, and it is under-recommended relative to how well it does.

Choosing not to sleep train

This is a complete answer and it deserves more than a sentence at the end of a methods list.

Plenty of families decide not to run any formal intervention. Some because it conflicts with how they want to parent, some because they tried and found it intolerable, some because their cultural or family context assumes close night-time contact, some because the baby's sleep is not actually the biggest problem in their lives right now. All of these are sufficient reasons and none of them require justification.

The evidence does not stand against this choice. There is no research demonstrating that infants who are not sleep trained have worse developmental outcomes. On the contrary — Pennestri and colleagues looked directly at whether uninterrupted infant sleep was associated with mental and psychomotor development at six and twelve months and found no significant association. Consolidated sleep arrives eventually for essentially everyone regardless of method, and the second-year and third-year data show waking declining across the board.

What not sleep training does mean is that the parental cost is real and needs managing rather than absorbing. If nobody is intervening in the baby's sleep, someone should be intervening in the adults': splitting nights properly, taking shifts, sleeping in a different room on alternate nights, accepting help, going to bed at 8:30pm without apology. Chronic sleep deprivation in adults is a genuine health issue and it does not become less so because the choice was deliberate.

The things that do help without any formal method: a consistent bedtime routine, an age-appropriate schedule, a dark room, and morning daylight. Those are available to everyone and none of them involve a protocol.

What the research shows, and what it cannot tell you

It is worth being precise here, because both camps overstate the evidence in opposite directions.

On effectiveness, the evidence is reasonably strong. A review commissioned by the American Academy of Sleep Medicine and published by Mindell and colleagues in Sleep in 2006 examined more than fifty studies of behavioural treatments for bedtime problems and night waking in infants and young children, and found that the large majority reported clinically significant improvement, with extinction-based approaches and parent education among the best supported. That is a real body of work.

On harm, the most-cited findings are the absence of detected harm. Gradisar and colleagues' 2016 randomised trial compared graduated extinction and bedtime fading against a control group and measured not only sleep outcomes but infant cortisol and, at twelve-month follow-up, emotional and behavioural outcomes and parent–child attachment — finding no significant differences. Price and colleagues, following up a randomised trial five years later in Pediatrics in 2012, assessed children at age six on emotional and behavioural problems, sleep, cortisol and the child–parent relationship, and found neither lasting harms nor lasting benefits.

Now the limitations, stated honestly. Sample sizes in the harm studies are small — Gradisar's trial enrolled a few dozen infants. Families in these trials are self-selected and motivated, which is not the whole population. Most outcomes rely on parent report. Follow-up periods are short by developmental standards, with five years being unusually long. The evidence in infants under six months is thin. And the outcome measures capture what researchers chose to measure — critics reasonably point out that absence of detected effect in the measured domains is not the same as absence of effect in all domains.

The fair summary: these methods work for a lot of families, and the studies that have looked for harm have not found it, within real limits on how hard they have looked.

Where the medical boundary sits

Three things fall outside what any behavioural approach can address, and they are worth separating out clearly before you start.

Safe sleep is not negotiable and is not a sleep-training topic. Whatever method you use or do not use, the American Academy of Pediatrics recommends that infants sleep on their back, on a firm flat surface, in their own sleep space free of soft bedding, bumpers, pillows and loose blankets, and that families room-share without bed-sharing for at least the first six months. This site does not attempt to summarise that guidance because it is not something to take second-hand — read it at the AAP and NICHD sources linked below.

Night feeding is a medical question, not a behavioural one. Whether your baby still needs calories overnight depends on their growth, intake and individual circumstances, and that is a conversation with your pediatrician. Do not night-wean on the strength of a chart from the internet, including this one.

And underlying causes need ruling out. Persistent night waking accompanied by snoring, mouth-breathing, observed pauses in breathing, unusual restlessness, poor weight gain, or evident pain warrants a medical evaluation. Sleep-disordered breathing in particular is meaningfully common in young children and does not respond to behavioural intervention — trying to sleep train through it is both futile and unfair to the child.

None of the above is a reason not to sleep train. It is the boundary around what sleep training is capable of doing.

Running whichever one you chose

The differences between methods matter less than the execution, and the execution failures are remarkably consistent across families.

The first is inconsistency between nights. Rotating between responses — checks on Monday, feeding on Tuesday, bed-sharing on Wednesday — is harder for a baby than any single approach, because there is no pattern available to learn. Whatever you pick, run it unmodified for at least a week before judging it.

The second is inconsistency between adults. If one parent is doing graduated checks and the other is picking up at the first sound, the baby is not receiving a method. Agree the specifics in advance, in daylight, in writing if necessary.

The third is quitting on night two or three, which is very often the worst night. Improvement in most protocols is not linear and the spike is expected. Decide in advance how many nights you are committing to, and hold that.

The fourth is starting with a broken schedule. An overtired baby fights sleep harder, not less, and no method compensates for wake windows that stopped fitting a month ago. Check the current numbers on our wake window calculator before you start, and see the homepage for the whole-schedule picture.

And finally: expect setbacks. Illness, travel, teeth and developmental leaps all interrupt, and a few nights of stepping back in is not the undoing of anything. Most families find that returning to the established pattern after a disruption takes a night or two, not a fresh start.

If a week of consistent application has produced nothing at all, stop rather than escalate. That is usually a signal that something else — the schedule, hunger, or a medical cause — is the actual bottleneck.

• Commit to one method, unmodified, for a minimum of seven nights
• Write down the exact protocol both adults will follow before night one
• Keep a simple log — down time, latency, wakings — because memory at 3am is not reliable
• Plan for night two or three to be worse than night one
• Keep bedtime routine, room darkness and schedule identical throughout
• Do not start during illness, travel, or the acute phase of a developmental change
• If a full consistent week produces no change at all, stop and look for a different cause