Sleep Training: Is it Bad?

Modified excerpt from Cribsheet below, on one of the hottest button topics: Cry it Out.

I posted about the best parenting advice a couple of weeks ago, and about 13% of it was in the category I call “Sleep-Related”. And a lot of this, in turn, amounted to a version of sleep train your child. By “sleep training” people typically mean some form of “crying it out” — broadly, any system where you leave the baby in his crib on his own at the start of the night, and sometimes let him fall back to sleep on his own if he wakes at some point during the night. The name refers to the fact that if you do this, your baby will cry some at the start of the program.

There are many versions of this: Ferber, Weissbluth, etc. All advocate slightly different approaches (do you check on them, how long do you wait before going in, whether you stay in the room, etc.) This approach is popular with parents, and advocated by many pediatricians. But it is also very controversial.

Go on the internet, and you’ll immediately find a variety of articles detailing the extensive long-term damage sleep training will do to your child. Google “cry it out,” and on the first page of results you’ll find an article by a PhD psychologist, Darcia Narvaez, entitled “Dangers of ‘Crying It Out’: Damaging children and their Relationships for the Long Term.” The article proceeds as you’d expect based on the title. It details the selfish reasons people would choose to do this, and the many long-term psychological issues it could create.

At its core, the concern from the opponents of “cry it out” is that your baby will feel abandoned and, as a result, struggle to form attachments to you, and ultimately to anyone else. It is worth a brief digression on where this idea comes from.

The answer: Romanian orphanages.

In the 1980s, a deep failure of reproductive policy left thousands of infants and children in Romanian orphanages. These children suffered all kinds of tragic deprivations, including limited food, as well as physical and sexual abuse. In addition, they had almost no adult contact as infants and children. They were left in their cribs for years with virtually no human contact, resulting in very late physical development, in addition to mental and psychological costs. Researchers who visited these children found the children could not form bonds with others, and many of them have struggled their whole lives.

This influenced the attachment-parenting philosophy, including views on the use of “cry it out.” One of the things visitors noticed in these places was the eerie quiet of the rooms the children were kept in. Infants and babies didn’t cry, because they knew no one would come. The argument is that “cry it out” is the same thing: Your baby will stop crying because she knows you will not come, just as the children in these orphanages did. And just as in those settings, her ability to attach to you and others will be forever changed.

This was a terrible and shameful episode that should never have happened. But it is also not comparable to the experience of most infants whose parents use “cry it out” methods. None of these suggest leaving the infant for months without any human contact, nor do they suggest subjecting children to the other types of physical and emotional abuse common in the Romanian orphanage experience.

Obviously the writers of anti-“cry it out” articles understand this, but in their view, “cry it out” is a continuum. The children left in these orphanages suffered extreme long-term consequences. Children who experience other types of chronic life stress—physical abuse, serious neglect—often have long-term problems. A few nights of sleep training probably will not do that, but who knows whether they endure smaller damages?

Fortunately, the literature does know—at least to some extent—and we can subject the question of whether sleep training is harmful to the data. But before getting into that, it seems useful to start with the basic question of whether sleep training works.

Does It Work?

Yes, this method works for improving sleep.

There are many, many studies on this, employing a variety of related procedures (many of these are randomized trials). A 2006 review covered nineteen studies of the unfortunately named “Extinction” method—the form of “cry it out” in which you leave and do not return—of which seventeen showed improvements in sleep. Another fourteen studies used “Graduated Extinction”—where you come in to check on the baby at increasingly lengthy intervals—and all showed improvements. A smaller number of studies covered “Extinction with Parental Presence”—in which you stay in the room but let the child cry—and these also showed positive effects.

These effects persist through six months or a year. This means that children who are sleep trained are sleeping better (on average) even a year after the training.

These methods do not completely solve all sleep problems from day one. And some children respond better than others, as do some parents. Not every baby who is sleep trained will sleep through the night every night, but they do sleep better on average.

The bottom line is that there is simply a tremendous amount of evidence suggesting that “cry it out” is an effective method of improving sleep.

It is worth noting that most of these studies—and, indeed, virtually all sleep books—recommend a “bedtime routine” as part of any sleep intervention. There isn’t much direct evidence on this—the review refers to it as a “common sense recommendation”—but it is generally included with all intervention approaches. The idea is to have some activities that signal to the baby that it is bedtime: putting on the baby’s pajamas, reading them a book, singing some kind of song, turning off the lights. Basically, no one recommends throwing a fully-clothed baby in the crib with the lights on, telling them it is bedtime, and closing the door.


While much of the popular discussion of sleep training focuses on its possible harms, much of the academic literature focuses on its possible benefits, including not only improvements in infant sleep, but also benefits to the parents.

Most important, sleep interventions seem to be very successful at reducing maternal depression. To take one example, an Australian study of 328 children randomized half into a sleep-training regime and the other half into a control group. Two and four months later, the authors found that the mothers of babies in the sleep-training arm were less likely to be depressed and more likely to have better physical health. They were less likely to use health services, as well.

This finding is consistent across studies. Sleep-training methods consistently improve parental mental health; this includes less depression, higher marital satisfaction, and lower parenting stress. In some cases the effects are very large. One small (non-randomized) study reported that 70 percent of mothers fit the criteria for clinical depression at study enrollment, and only 10 percent after the intervention.

Obviously we want to think carefully about any possible risks to babies, but the fact that sleep training is good for parents should not be ignored. And sleep is also beneficial to development for babies and kids. Settling into a good sleep routine—one that will ensure longer and higher-quality sleep—could have long-term positive effects for children.

Is “Cry It Out” Harmful?

“Cry it out” works, helps parents and kids sleep better, and improves at least parental mood. Is it harmful for your child?

There are a number of good randomized trials which speak to this. One representative study from Sweden, published in 2004, took ninety-five families and randomized them into a sleep-training regime involving a form of “cry it out.” The authors focused on whether behavior during the day was impacted by the nighttime—basically, they asked whether the infants were less attached to their parents during the day as a result of being left to cry during the night.

This particular study found that, in fact, infant security and attachment seemed to increase after the “cry it out” intervention. It also found improvements in daytime behavior and eating as reported by the babies’ parents. Note that this is the opposite of the concerns raised about “cry it out” methods.

This study is not alone. A 2006 review of sleep-training studies, which included thirteen different interventions, noted the following: “Adverse secondary effects as the result of participating in behaviorally based sleep programs were not identified in any of the studies. On the contrary, infants who participated in sleep interventions were found to be more secure, predictable, less irritable, and to cry and fuss less following treatment.”

This evidence focuses on immediate impacts on the infant. But this isn’t necessarily the main concern among those who shun “cry it out.” Instead, the worry is about longer-term impacts. Yes, the infant cries less—maybe even less during the day—but because they have given up, not because they are happier.

To more fully address this, we need to follow sleep-trained children to older ages to see whether there are long-term risks. This adds to the difficulty of running a randomized trial, of course, since longer-term follow-up is both difficult and expensive. However, we do have one example.

This study was run in Australia, with 328 families recruited when their babies were eight months old. The authors first showed that the intervention improved sleep and lowered parental depression. But they didn’t stop there. They returned to evaluate the children a year later and, most notable, five years later, when the children were almost six. In this later follow-up, which included a subset of the original families, the researchers found no difference in any outcomes, including emotional stability and conduct behavior, stress, parent-child closeness, conflict, parent-child attachment, or attachment in general. Basically, the kids who were sleep trained looked exactly like those who were not.

This study—as well as the others and various review articles—do not point to either long- or short-term harms from “cry it out.” And it works, and it is good for parents. This paints a pretty pro-“cry it out” picture. But it is not one that everyone agrees with.

A number of academic articles argue against “cry it out” from a theoretical perspective. One good example comes from an article published in 2011 in a journal called Sleep Medicine Reviews. The authors of this article presented a case against “cry it out,” largely based on the idea that infant crying is intended as a signal of distress, and parents should therefore not be encouraged to ignore it. They draw on the attachment theories (i.e., the orphanage literature), and argue that parents who engage in this are ignoring their children’s efforts to begin communication with them.

The fact that “cry it out” works is not compelling to these researchers and, indeed, is an indication of harm. As one article puts it, “Is the cessation of crying a ‘cure’ or is it that the child has ‘given up’ and is now depressed and has partially withdrawn from the attachment dyad?”

The primary argument offered by this and similar papers is that infant crying is a signal of stress (probably true) and that stress, even over a short period of days or weeks, may have long-term consequences for babies (speculative). These authors often point to one particular study to support these stress claims. That study, published in 2012, followed twenty-five infants and their mothers in New Zealand over a five-day inpatient treatment in a sleep lab. The goal of the stay in the lab was to sleep train the infants. Nurses in the study collected data on the stress hormone cortisol in both the babies and their mothers, and were also responsible for putting the infants to sleep, and monitoring the sleep training.

Before the sleep training each day, the babies’ and moms’ cortisol levels were tested and recorded.  This was done again after the infant fell asleep. On the first day, the babies all cried. Their cortisol levels were the same before the training and after they fell asleep. Their mothers’ cortisol levels were also the same before the babies cried and after they were asleep. This was the same on the second day.

On the third day, none of the infants cried (see above: sleep training works). However, they showed the same cortisol patterns: equal before bedtime and after they fell asleep. But for the moms, this changed: they had lower cortisol levels in the later period, when the babies weren’t crying.

The authors suggested that this presents a problem with sleep training. In particular, they note that after sleep training, the mother’s stress levels do not stay in sync with the infant’s, which they interpret as possible evidence that the attachment between mother and infant is weakening.

A number of commentators have argued that this is an over-interpretation of the study. For one thing, there is no baseline level of cortisol given, so we actually have no way to know if the babies were even experiencing elevated stress. For another, the study stopped after three days (or at least the data reporting did), so we don’t know what happened later.

But even beyond this, it is unclear why differing levels of cortisol for moms and infants after sleep training is a problem. Effectively, this study shows that mothers are more relaxed after sleep training occurs, and that there are no other changes for the infant. This seems like a positive result, not a negative one.

Fundamentally, the argument against sleep training is theoretical. We know that abuse and neglect have long-term consequences, so how can we be sure that four days of a baby crying itself to sleep doesn’t? You might think you could look at the data on long-term impacts and note that everything seems fine, but the theoretical counter-argument is simply that for some children, this is devastating, and you do not know who those children are.

This argument is nearly impossible to refute. There is no way to prove or disprove it. You’d need a huge sample size, and even then most studies wouldn’t be designed to pick up this kind of heterogeneity.

A related argument is that although children may look fine at five or six years old, the damage from sleep training may not manifest until they are adults. Again, very hard to study.

I think it is fair to say that it would be good to have more data—it’s always good to have more data! And yes, it is possible that if we had more data, we would find some small negative effects. The studies we have are not perfect.

However, the idea that this uncertainty should lead us to avoid sleep training is flawed. Among other things, you could easily argue the opposite: maybe sleep training is very good for some kids—they really need the uninterrupted sleep—and there is a risk of damaging your child by not sleep training. There isn’t anything in the data which shows this, but there is similarly nothing to show that sleep training is bad.

You could also argue that the effects of maternal depression on children are long-lasting, and therefore this intervention may have beneficial long-term effects. This seems in many ways more plausible.

You’ll have to make a choice about this without perfect data. (This is true of virtually all parenting choices. Blame the parenting researchers!) But it would be a mistake to say, for example, that not sleep training is the “safest option.”

Does all this mean you should definitely sleep train? Of course not—every family is different, and you may really not want to let your baby “cry it out.” You need to make your own choices, just as with everything else. But if you do want sleep train, you should not feel shame or discomfort about that decision. The data, imperfect as it is, is on your side.

Recommended Posts You Might have Missed

If you liked this post, here are are a few older once that might tickle your fancy….

Weigh in or Share!

Keep the thoughts coming.

Leave a comment

Share ParentData

Write to Me!