Is sleep training safe? What the research actually found

Quick answer:

• Two randomized trials looked for harm to attachment and stress. Neither found any.
• The same trials found the benefits faded too. That is reassuring about long-term outcomes, not proof of absolute safety or of a lasting fix.
• Both studied babies from about six months. Below that, the advice is convention, not trial data.
• Crying isn't required for an intervention to work: in one randomized trial, bedtime fading with no extinction substantially shortened how long babies took to fall asleep.

If you've read anything about sleep training, you've found flatly opposite advice. It's harmful. It's necessary. It fixed everything in three nights. The short honest answer is narrower than either camp wants: in the babies who have been studied, nobody has been able to detect harm, and nobody has been able to detect a lasting benefit either. I'm writing this as someone who went through most of the list below with my eldest.

🌙 Map out the day: enter your baby's age in the sleep schedule calculator for naps and a bedtime that fit their rhythm.

Quick reference: sleep training basics

For the practical schedule half of this, see baby sleep guide 0–2 years.

Why this feels so confusing

Most parents go looking for the right method. Sleep rarely works that way. Two babies can respond completely differently to the same approach, because the method usually isn't the thing driving the night.

What is usually driving it:

• timing: wake windows, bedtime, nap length
• overtiredness that has been building for days
• night-to-night inconsistency
• developmental change that has quietly wrecked the old schedule

When those are unclear, any method can look like it worked, or failed. That is most of the reason the internet contradicts itself. The same approach "saved one family" and "did nothing for another" because the underlying pattern was different in each house.

What "sleep training" actually means

Not one method. A range, and the range matters, because the research covers some of it and not other parts of it.

• Schedule only. Wake windows, a predictable bedtime routine, consistent timing. No named method at all.
• Bedtime fading. Move bedtime later to meet the time your baby is actually falling asleep, then walk it back. No extinction involved.
• Gentle presence methods. Gradual retreat, pick-up-put-down, the chair method.
• Graduated extinction. Checking at set intervals. This is what most people mean by "sleep training".
• Full extinction. No checks. Its effectiveness has support in the older literature, but the two trials carrying most of the long-term outcome data did not test this version directly.

The word covers everything from "we made dinner-bath-book reliable" to "we did cry-it-out for three nights". Those are different experiences with very different evidence behind them.

Approaches at a glance

The speeds below are our practical convention from working with schedules, not trial results. Trials measured sleep latency and night wakings, not "how many weeks until it feels fixed".

What the research actually found

The effectiveness literature is far wider than this argument makes it look. A 2026 systematic review gathered 59 such studies in children under five. But what those studies measure is almost always night-time sleep, mostly through parent questionnaires and diaries; attachment barely appears among the assessed outcomes at all (Pétrin et al., 2026). When the question is specifically about stress, attachment and long-term harm, there is far less direct data. That is why two randomized trials matter so much here, and why it's worth knowing exactly what each one did.

Price et al., 2012, *Pediatrics*. 326 children whose parents reported a sleep problem at 7 months. Behavioural techniques were delivered by nurses over one to three consultations at 8 to 10 months, against usual care. The children were assessed again at age 6, including emotional and conduct scores, sleep, stress regulation, the child-parent relationship, and attachment measures. The result was nothing, in either direction. No differences on any outcome. The authors' own conclusion is that behavioural sleep techniques have "no marked long-lasting effects (positive or negative)", and can be used to reduce the short to medium term burden.

Read that twice, because it cuts both ways. The scary version of sleep training didn't show up at age 6. Neither did the sales pitch.

Gradisar et al., 2016, *Pediatrics*. Smaller and more targeted: 43 infants aged 6 to 16 months, randomized to graduated extinction, bedtime fading, or a sleep education control. This trial went looking for stress physiology specifically. Salivary cortisol was sampled, and instead of rising in the intervention groups it declined slightly. Twelve months later the mother-child pairs went through the strange situation procedure, the standard laboratory attachment assessment, and there were no differences in secure versus insecure attachment between groups.

The bedtime fading arm is the part most articles skip. It improved sleep latency substantially, without any extinction. The reductions in night wakings and time awake during the night were larger for graduated extinction, so the two are not interchangeable. But for getting a baby to sleep faster, extinction was not required.

Where the evidence stops

Being straight about the limits is the only way the reassurance is worth anything:

• Sample sizes are modest. Gradisar included only 43 infants. A sample that size can pick up larger differences, but it is too small to reliably rule out small or rare harms.
• Neither trial studied young infants. Price enrolled at 7 months, Gradisar from 6. Anything said about a four-month-old is extrapolation.
• The methods studied were the moderate ones. Nurse-delivered behavioural advice and graduated extinction. Not unsupported full extinction in a very young baby.
• Follow-up is years, not decades. Age 6 is a real follow-up. It is not a lifetime.

A review of behavioural treatments commissioned by the American Academy of Sleep Medicine reached the same broad position: these interventions produce reliable short-term improvement in bedtime resistance and night waking (Mindell et al., 2006).

What matters more than the method

1. Age. Around six months is where the evidence begins. Younger babies often genuinely need night feeds, and the NHS guidance for small babies is about routine and settling, not training (NHS).
2. Timing. Wake windows, bedtime and nap length usually matter more than the label on the method. A bad bedtime sinks every approach. See signs your baby is overtired.
3. Consistency. Hold whatever you choose for one to two weeks. Changing method every third night is the single most common way this fails.
4. Whether you can actually sustain it. A method you can't get through emotionally at 2 a.m. isn't the right method, regardless of what the evidence says about it.
5. Your own state. Sleep deprivation has measurable effects on mood and judgement. That is not separate from what's good for the baby.

For the schedule half, baby schedule by age 0–2 years covers wake windows and bedtime targets.

When to wait

Pause, or work on the basics instead, if:

• your baby is under 4 months
• there's acute illness or a big teething peak running the nights
• you're inside a developmental leap: rolling, crawling, walking, a language explosion
• days swing wildly, so there's no baseline to change

Sometimes the right answer is "wait two weeks until this passes". That is judgement, not failure.

For what a regression looks like, see sleep regression: what actually helps.

Start here, before you choose anything

1. Track wake windows, nap length and bedtime for five to seven days.
2. Look at the best nights. What was different? Earlier bedtime, longer nap, calmer evening?
3. Look at the worst nights. Same question.
4. Change what the pattern points at, which is usually 30 minutes off bedtime or a cap on one nap, not a named method.
5. If the pattern is clean and settling is still hard, then consider a structured approach.

Most families who are sure they need sleep training need half an hour off bedtime first.

For the mistakes that get misread as method failures, see the biggest baby sleep mistakes parents make.

A balanced position

Sleep training is not required. It also has not been shown to be harmful in the age range and with the methods that have actually been tested. Both halves of that sentence are load-bearing, and most articles only carry one of them.

With my eldest I tried everything I could think of. More sleep, less sleep, a different place to sleep, blackout, white noise, no white noise, letting her cry and not letting her cry. If someone had told me back then that howling at the moon on one leg helped, I was desperate and tired enough to have tried that too.

What I'm left with is the impression that in some phases nothing helped. Not even the hardest thing. The text above says to hold a method for one to two weeks; I was changing them every few nights, because I didn't have the capacity for anything else. I'm not offering that as advice. I'm offering it as data: even when you know what the literature says, the phase where nothing works still exists.

It's worth saying plainly that this is a topic where people's convictions run well ahead of the data on both sides. The trials are small, the follow-up is finite, and the strongest honest claim available is that nobody looking for damage has found it. If that isn't enough certainty for your family, that is a completely reasonable place to land, and skipping it costs your child nothing the research can measure.

Frequently asked questions

Is sleep training harmful to babies?
No harm has been detected in the trials that looked. Price et al. followed 326 children to age 6 and found no differences in emotional or conduct scores, stress regulation, the child-parent relationship, or attachment. That is the strongest available answer, and it is an absence of evidence of harm rather than proof of safety in every situation.

At what age can you sleep train?
The trial evidence starts around six months; Price enrolled families whose babies had sleep problems at 7 months. The common "4 to 6 months" advice is practice convention. Under 4 months, work on schedule and routine only.

Does sleep training cause attachment problems?
Both trials looked at parent-child relationship outcomes, but not in the same way. Gradisar assessed attachment directly, using the strange situation procedure a year after the intervention, and found no difference between groups. Price found no differences in the parent-child relationship outcomes it measured at age 6. Either way, what happens across your whole day matters more to attachment than what happens at bedtime.

Do I have to let my baby cry?
No. In Gradisar's trial the bedtime fading group, which involves no extinction, improved substantially on how long babies took to fall asleep. Plenty of families get where they're going on timing and routine alone.

What's the gentlest approach?
Schedule and routine consistency with no method at all, then bedtime fading, then gradual-presence methods. Many families never need to go past the first one.

What if it doesn't work?
After one to two consistent weeks with no movement, check the timing before you change method: wake windows, bedtime, nap length. Then check whether something is going on underneath, like teething, illness or a leap. Sometimes the answer is to stop and try again in a month.

How KidyGrow can help

The hardest question in this whole subject isn't which method to pick. It's whether last night was the method failing, or the day sabotaging it: the 30-minute nap, the late bedtime, the afternoon at the shops.

KidyGrow remembers the days you can't hold in your head. Day one it knows nothing about your baby, so the plan is age-based. By day eight it has your own week, and the daily brief can say something specific: the two hardest settles this week both followed a nap under 40 minutes, and the two easiest both followed a bath before 6:30. That's not a verdict on any method. It's the piece of information you needed before choosing one.

It takes three to five days of logging to say anything that concrete. Some weeks it finds nothing, because the week was a cold and a growth spurt, and no pattern survives that.

For the wider playbook, see baby sleep guide 0–2 years.

_This content is educational and does not replace professional sleep or medical advice. If you're struggling significantly, talk to your pediatrician._

Sources

1. Price AMH et al., 2012, Pediatrics - Five-year follow-up of harms and benefits of behavioral infant sleep intervention: randomized trial
2. Gradisar M et al., 2016, Pediatrics - Behavioral interventions for infant sleep problems: a randomized controlled trial
3. Pétrin M et al., 2026, Sleep Medicine - Behavioral sleep interventions in children aged 0-5 years: a systematic review of content, delivery, and 24-hour sleep outcomes
4. Mindell JA et al., 2006, Sleep (American Academy of Sleep Medicine review) - Behavioral treatment of bedtime problems and night wakings in infants and young children
5. NHS - Helping your baby to sleep
6. AAP HealthyChildren: Healthy Sleep Habits, How Many Hours Does Your Child Need? (accessed 2026).