A carousel is currently circulating a major assertion that sleep training can turn out to be the modern equivalent of spanking. The reasoning is that spanking was once a totally normal practice; experts used to advise it, devoted parents used to back it up, and people accepted it because "it works.” As evidence of its harm built up, the advice changed, and now most of us look back and wonder how it could have ever been standard guidance.

The carousel claims sleep training can follow the same logic. Babies never truly learn to sleep on their own; instead, they stop crying because they realize no one is coming, even though their nervous systems remain disturbed and enter a "shutdown" state. The argument continues that one day we will look back on the practice of leaving babies to cry with the same disbelief that so many of us currently feel about spanking.

I can see why this comparison has such an impact. It's emotionally compelling because it leaves many thoughtful parents feeling uneasy. Many of you have also sent this argument to me directly. I want to take this matter seriously, since the fundamental question is whether this constitutes a fair comparison.

Science is always evolving. Medicine has at times made mistakes. Parenting advice has changed as our knowledge of child development has increased and will continue to do so. It must be clear that just because we haven't proven something causes harm doesn't mean it can't. There is a difference between a lack of evidence and evidence that something is not present, which is why the carousel is right to highlight this point.

And being mistaken about one parenting method in the past does not automatically mean we are mistaken about another. Each practice has to be evaluated based on the evidence we have for that specific practice. That is the step this carousel skips, and I think it is the most important one.

I understand why the comparison lands. Both spanking and cry-based sleep training can involve a distressed or crying child. Both result in a change in behavior. And some version of "but it works" has been used to defend both practices.

They also raise a question I think about often as a pediatrician: does stopping a behavior mean we have actually helped the child? That is a fair and important question, and one I never want us to stop asking.

But this is also where the comparison starts to break down. With spanking, causing physical pain or discomfort is the intervention. With cry-based sleep training, crying can happen in response to a change in how a parent responds at bedtime. The presence of crying does not make the mechanisms the same. 

And even if two practices share some surface similarities, that does not mean they produce the same outcomes. That question has to be answered by clinical experience and research, not by analogy.

We did not stop recommending spanking because children cried when they were spanked. We stopped because evidence of harm built up.

Over decades, research linked physical punishment with increased aggression and behavioral problems, poorer mental health outcomes, and strain in the parent-child relationship. Eventually, the evidence was strong enough that major professional organizations, including the American Academy of Pediatrics, advised against it.

That history matters. The question is whether we are seeing a similar pattern of evidence with sleep training.

What kind of results would I see if sleep training were on the same level as spanking? It wouldn't just be that the babies cried; crying is the starting point of the theory, not evidence of it. I would expect to find signs in the results that matter over time:

If the idea is that the baby has worked out “nobody is going to respond to me, so I should stop sending out signals”, we should eventually be able to observe the effects of that belief somewhere other than the fact that the baby stopped crying at bedtime. So what do the researchers find when they look?

A 2016 study by Gradisar and colleagues, published in Pediatrics, included 43 infants aged 6 to 16 months who received one of three interventions: graduated extinction, bedtime fading, or sleep education. The researchers assessed sleep, salivary cortisol levels, maternal stress, emotional and behavioral outcomes, and attachment. The groups that received the behavioral interventions had better sleep. Cortisol levels showed small to moderate decreases compared with the control group, contrary to what a "nervous system in shutdown" theory would expect. At 12 months, there were no significant differences in attachment or in the emotional and behavioral outcomes. The honest limitation is that this was a very small study. Forty-three infants are not enough to settle this question, and I would by no means pretend that they are.

A five-year follow-up study by Price and colleagues, published in Pediatrics in 2012, examined children six years after they participated in a randomized behavioral sleep intervention. The researchers assessed the children's emotional and conduct problems, psychosocial functioning, stress regulation, the child-parent relationship, attachment-related outcomes, and the mother's mental health. They found no meaningful differences between the intervention and control groups. As with the previous study, limitations included family attrition over time, a focus on specific interventions within a particular population, and the inability to establish that every kind of sleep training is harmless for every child.

A 2006 review in the journal Sleep by Mindell and colleagues examined 52 treatment studies of behavioral interventions for bedtime difficulties and night wakings. They found strong evidence that these approaches can improve sleep. However, the key limitation is that demonstrating effectiveness does not establish long-term safety. While these studies provide more information on sleep improvements, they do not address possible subtle developmental effects that may emerge many years later.

A 2020 systematic review in Acta Paediatrica found modest short-term benefits of behavioral interventions but also identified significant evidence gaps—specifically, research on infants under six months of age, high-risk groups, diverse cultures, and the effects of extinction-based methods. The broader literature shares similar limitations, including reliance on parent-reported outcomes, mixed-age participants, variation across interventions, and several potential confounding variables.

More on sleep-training evidence in this highly downloaded podcast episode with Sleep Medicine Physician Dr. Sujay Kansagra or this newsletter from the archives.

What then are we to conclude? The research does not allow us to state that "We have definitively proven that every form of sleep training is harmless." However, it also does not permit anyone to assert that "We know that babies stop crying because they have learned that nobody is coming, their nervous systems shut down, and as a result their attachment is damaged." That conclusion goes well beyond what the evidence shows. When researchers have specifically looked for the harms mentioned by the carousel—regarding attachment, emotional health, behavior, and stress regulation—they have not found any.

I would also like us to continue studying sleep training. We need larger studies, longer follow-up, more diverse populations, and better ways of measuring outcomes. If strong evidence eventually shows that sleep training causes harm, I will change my recommendation. I have changed my mind as new evidence has emerged before, and I will always have the humility to keep learning and do so again.

But we also need to be realistic about how difficult it is to design the "perfect" sleep training study.

Imagine the ideal experiment. You would take thousands of babies who were identical in every important way, randomly assign them to different sleep methods, make sure every family followed the assigned method exactly, and then follow those children for twenty or thirty years while somehow keeping everything else about their lives the same.

Families are not laboratory conditions, and thank goodness for that.

A parent-child relationship is built through thousands of interactions throughout the day. How does a parent respond when their child cries, gets hurt, needs comfort, separates from them, reunites with them, makes a mistake, has a tantrum, or simply wants connection? Two families could use the exact same sleep training method and have very different relationships with their children outside of bedtime.

And the list of other variables is enormous: infant temperament, baseline sleep patterns, feeding, parental mental health, caregiver responsiveness, socioeconomic status, parental leave, number of children, cultural expectations around sleep, room-sharing or bed-sharing, family stress, how consistently the method was used, how much crying occurred, the baby's age, and medical or developmental differences.

There is also self-selection. Families who choose to sleep train or sleep training with a cry method may already differ from families who do not. They may be more comfortable with crying. Their babies may have different temperaments. One group may be more exhausted or anxious. Parents who choose different sleep approaches may also parent differently in a hundred other ways.

Randomization helps address some of this, but long-term randomized trials get messy. Control families may eventually sleep train. Intervention families may stop. Families drop out of studies. Children grow. Circumstances change. Life happens.

All of this makes it incredibly difficult to isolate one parenting practice and say with certainty that it caused a specific outcome years later. That does not mean we stop studying it. It means we understand what the research can and cannot tell us.

Every study has limitations. That applies to studies suggesting benefit or finding no evidence of harm, and it applies just as much to studies and theories used to argue that sleep training causes harm. We should scrutinize them all equally.

This is not a randomized controlled trial, and I don't want to pretend clinical experience can answer a research question. But something unique about pediatrics gets lost in these conversations: we get to watch children grow up.

I have cared for children from infancy through toddlerhood, preschool, elementary school, and beyond. I know many of these families well. I know their histories. I know which families sleep-trained and which did not.

Over years of visits, I am watching much more than whether a child sleeps through the night. Does this child turn to their caregiver for comfort when scared or hurt? Can their caregiver comfort them? Is there warmth and affection between them? How does the child handle separation and reunion? How do they communicate their needs and manage frustration? How does their emotional regulation develop? Are behavioral or developmental concerns emerging over time? I also see how parents respond during difficult moments. Do they use nurturing, responsive discipline? Do they reconnect after conflict? Is crying generally met with comfort and care?

That broader context matters enormously.

A loving, responsive parent who decides to use a cry-based sleep method for a specific period at bedtime is very different from a caregiver who routinely ignores a child's distress throughout the day, uses isolation as punishment, yells, hits, or is consistently emotionally unavailable. Simply saying that a child was "left to cry" strips away all of that context.

And this is part of why, clinically, I do not see appropriately used cry-based sleep training within an otherwise loving, responsive relationship as inherently harmful. I have followed many of these children for years. I see them run into those same parents' arms when they get hurt, seek them when they are scared, protest separation, calm when reunited, and develop warm relationships with them.

Of course, these observations are not formal attachment measurements, and I cannot diagnose "secure attachment" during a well visit. But pediatric care happens over years, not one visit. And over those years, I have not observed a pattern of social-emotional or parent-child relationship concerns associated with cry-based sleep training

I know the obvious response: "Couldn't a pediatrician once have said the same thing about spanking? Plenty of kids who were spanked seemed fine."

Yes. That is a fair criticism, and it is exactly why clinical experience cannot prove that sleep training is harmless. I can miss subtle effects. I have my own biases. I do not have a control group in my waiting room, and population-level differences can be difficult to see when caring for one child at a time.

A baby crying less after sleep training could have several explanations. They may have gotten better at falling asleep independently. They may have learned the predictable rhythm of bedtime. They may feel less distressed as the routine becomes familiar. They may have learned that crying does not change what happens at bedtime. Or it may be some combination of these.

But if we are going to assign an internal thought to that baby, why does it automatically have to be, "Nobody is coming, so there is no point in crying"? Why couldn't it equally be, "I know this routine. I'm safe. I'm tired. I can fall asleep"?

The answer is that we cannot know either from the crying behavior alone. A baby stopping crying tells us that the behavior changed. It does not tell us what the baby is thinking or feeling.

So "they stopped crying because they learned nobody is coming" is a theory, not a measured outcome. The same applies to claims that the baby's nervous system has entered "shutdown." That is a physiological claim, and it requires physiological evidence.

The evidence we do have does not support that conclusion. The Gradisar trial found small-to-moderate declines in salivary cortisol among the behavioral intervention groups compared with controls. Is that study small? Yes. Does cortisol have limitations as a measure? Also yes. But we cannot dismiss cortisol as an imperfect measure when the finding does not support a theory while treating "nervous system shutdown" as established physiology without measuring it. The same rules of evidence have to apply to everyone, including me.

Babies and children experience distress. Separation can cause distress. Vaccines can cause distress. Being buckled into a car seat can cause distress. Starting daycare can cause distress. A parent leaving the room can cause distress.

None of that means we should ignore distress. Our children's tears deserve our attention, always. But crying alone cannot tell us whether an experience is harmful. To answer that, we need context, duration, developmental stage, the quality of the caregiver relationship, and ultimately, outcomes. I wrote about this same idea in my newsletter on crying and trauma, and in a PedsDocTalk podcast conversation on what actually causes childhood trauma, and it applies here just as much: a hard moment and a harmful one are not always the same.

One more piece of nuance gets flattened in these debates: sleep training is an umbrella term for changing how we support sleep. That can include bedtime fading, gradually reducing parental help, timed check-ins, camping out in the room, and yes, methods that involve more crying. Lumping all of that together as "leaving babies alone to cry" simply isn't accurate.

As a pediatrician, I learn a tremendous amount from families. I can recommend something based on the evidence and my clinical experience, and then a family comes back and tells me what actually happened in their home. I learn from the families who say something worked, but I learn FAR more from the families who tell me it didn't.

Earlier in my career, I was very much a graduated-extinction/Ferber person. Then I met more families. I met the baby who vomited when they cried hard, and I thought, okay, a cry method may be too much for this child. I met families who tried a cry method, hated it, stopped, regrouped, and did beautifully with something more gradual. Those families taught me to learn other approaches and become much more flexible in how I talk about sleep. Through my work as a pediatrician, I have also become more open to conversations about safer bed-sharing. Not because I think every family should bed-share, but because I have learned the importance of meeting families where they are and giving them information that reflects what they are actually doing.

But I have also seen the other side. I have cared for parents so sleep deprived they were barely functioning. They tried the gradual approaches. They tried more parental involvement. They were exhausted, overwhelmed, and desperate for sleep. Some eventually chose a cry-based method, came back functioning again, and asked me, "Why didn't we do this sooner?"

When I first started my podcast, I recorded an episode with moms from this community sharing their own approaches to sleep. Some changed methods along the way. Some realized what worked for one child did not work for another. Some chose sleep training, while others didn't. Years later, it is still the most downloaded episode of my podcast.

I think it resonated because it showed what I have seen clinically for years: families are different, children are different, and sometimes the right decision changes as you learn more about the child in front of you.

I have seen too many versions of this to believe there should be one blanket recommendation for every family. But I have also seen too much to be comfortable vilifying an option that can genuinely help some families.

Age matters. Temperament matters. Feeding, growth, and medical history matter. The method matters. How a child responds matters. Parental mental health and comfort matter. Family culture and values matter.

Supporting sleep training does not mean recommending it to every family. It means giving families accurate information about their options, helping them choose an approach that makes sense for their child and their circumstances, and changing course when something is clearly not working.

We can discuss other approaches to sleep without vilifying cry-based methods. And we can acknowledge that cry-based methods help some families without pretending they are right for every child. Parents deserve the full menu of options, not fear around one of them.

I take the comparison to spanking personally. I was spanked as a child, and I can see ways that approach to discipline affected me. It is part of what pushed me to learn so much more about child psychology, attachment, discipline, and the parent-child relationship as an adult and as a pediatrician.

But there is an important difference between spanking and sleep training that I think this comparison misses.

Spanking is a discipline method. The physical pain or fear is used intentionally in response to a child's behavior, with the goal of changing that behavior. And when spanking is a family's primary approach to discipline, it is not necessarily one isolated moment. It can become part of the repeated way a child experiences mistakes, conflict, limit-setting, and big emotions.

Sleep training is not discipline. A parent is not using crying, fear, or physical pain to punish a baby for waking up. The goal is to change how the parent supports the child in falling asleep. With some methods, crying happens during that process. That does not automatically make the crying meaningless or mean every method is appropriate for every child, but it does make the mechanism fundamentally different from spanking.

And importantly, a family can use a cry-based sleep method at bedtime while being highly responsive throughout the rest of the child's day. They can comfort their child when they are hurt, respond to their emotions, use nurturing discipline, repair after conflict, and have a warm and connected relationship. The fact that a parent did not immediately respond to every cry during a specific sleep intervention does not tell us how that parent responds to their child across the thousands of other interactions that shape their relationship.

If sleep training eventually follows the same path as spanking, the evidence should take us there. We should not decide the destination first and work backward.

You do not have to sleep-train. You do not have to use a cry-based sleep approach if that is not something you’re comfortable with or to be fair.  Responsive settling, gradual approaches, and simply riding out the early months are all legitimate paths, but so are cry methods and sleep-training in general for many families.

And if you did sleep train, you do not need to be frightened into believing you harmed your child's brain, nervous system, or attachment. The evidence we have does not support that claim, and a viral carousel is no substitute for it.

We still don't know everything. Maybe research 20 years from now will change my mind, and if it does, I will definitely have the humility to say so out loud in this newsletter to all of you. Until then, I will use the same standard I want applied to every parenting recommendation: What does the best evidence we have actually show? What does my clinical experience show me? Not what we imagine might be happening. Not what makes the most compelling Instagram carousel. What we can actually support with evidence and experience. 

Because you deserve information, not fear, and your baby deserves parents who feel steady in their choices, no matter how your family gets to sleep.  For more on sleep, check out this newsletter with all of my sleep resources.

I would love to hear from you on this one. Did you sleep train, choose a gradual approach, or skip it entirely? How did the noise online affect how you felt about your decision? Hit reply and tell me. 

Bringing home a newborn comes with a flood of questions, and most of them are completely normal to have. New parents deserve clear answers, not more anxiety. Setting boundaries with visitors is not rude, it is protective. And so many of the things that look alarming on a newborn's skin are actually harmless and expected. When parents know what is normal, they can spend less time worrying and more time bonding with their baby!

I discuss: