Is It Dangerous to Wake a Sleepwalker?

01What to do in the next few minutes

Steer, do not stop. The instinct is to plant yourself in front of them and block the way, and that is the one thing to avoid, because a body in the path is an obstacle to be pushed through rather than a person to be obeyed. Stand slightly to one side. Put a light hand at the shoulder or the small of the back and apply direction rather than resistance, the way you would guide somebody through a crowd. The NHS advice is exactly this, to try to gently guide them back to bed.

Keep your voice low, short and repetitive. Their name, then something simple and true, then their name again. Do not ask questions, do not argue with whatever they say, and do not try to explain to them where they are, because that explanation is not landing the way you think it is. If they say something that makes no sense, agreeing with it and continuing to steer works considerably better than correcting it.

Give yourself light, not them. A lamp in the next room or a hall light so that you can see the floor, the step and the corner of the coffee table. A ceiling light switched on directly over somebody in this state is a startle, and startling is the thing you are trying to avoid.

Then let it finish. If you get them back to a bed, they will usually settle on their own. MedlinePlus notes that if they are not disturbed, sleepwalkers go back to sleep, though they may fall asleep somewhere different or slightly odd, which is not an emergency and does not need fixing at two in the morning. Most episodes are brief. MedlinePlus puts the majority at under ten minutes, and the Mayo Clinic describes a typical bout as several minutes, arriving one to two hours after falling asleep, which is why so many households recognize the same window night after night.

One last thing, and it is the easiest to forget: check the exits before you go back to bed yourself, not the next morning.

02Whether waking them can hurt them

No. That is the whole answer, and it is worth stating without hedging, because the hedged version is what leaves people frozen.

There is no described mechanism by which being woken damages a sleepwalker, and none of the clinical bodies that publish on this describe one. What they describe instead is much more ordinary. The NHS says that somebody woken suddenly will be confused, will not know what is happening, and may be scared, angry or upset. MedlinePlus says confusion or disorientation for a short time. That is the actual cost of waking somebody: a few unhappy minutes, and possibly some crossness with you in the morning.

What does hurt sleepwalkers is the house. MedlinePlus makes the point directly, that sleepwalkers are commonly injured when they trip and lose their balance, and Mayo's list of complications is furniture, stairs, wandering outdoors and, rarely, injuring somebody nearby. In a 2008 study of forty adult sleepwalkers, Zadra and colleagues open by noting that sleepwalking is one of the leading causes of sleep related injury. None of that risk comes from a hand on a shoulder. It comes from a person moving through a room they cannot properly see.

So, the thing everybody has heard is pointed at the wrong hazard. It is not that touching them is dangerous. It is that a staircase is.

03Why the guidance still says guide rather than wake

If waking is harmless, it is fair to ask why every clinical source still leads with guiding. The instruction is real: a 2023 review of diagnosis and management of these conditions by Mainieri and colleagues in Diagnostics lists, among the standard measures, instructing parents and bed partners to refrain from interacting with or trying to wake the person, alongside specific instructions for making the environment safe.

The reason is not danger. It is that the attempt often does not do what you want it to do.

This has been measured directly, and the study answers the question better than any advice does. Pilon and colleagues, writing in Neurology in 2008, played sounds to ten adult sleepwalkers and ten people without the condition during deep, slow wave sleep, once on a normal night and once during recovery sleep after twenty five hours awake. In the control group, not a single sleepwalking episode was induced by anything they did. In the sleepwalkers, the sounds triggered episodes: thirty percent of them had at least one during normal deep sleep, and after the sleep deprivation, every single one did.

Then comes the detail that explains the hallway. The researchers found no significant difference between the volume of sound that induced an episode and the volume that simply woke somebody up. In other words, the same nudge that would rouse an ordinary sleeper can, in somebody predisposed to this, start an episode instead of ending one. You are not choosing between waking them and leaving them. You are choosing between two outcomes you do not control, and guiding is the option that does not put that coin in the air.

The other half of the folklore is worth clearing away here too, because it is the part that makes people give up too early. Sleepwalkers are not unrousable. Pilon and colleagues tested that in 2012 and found they needed a louder sound than controls only during REM sleep. In deep slow wave sleep, the stage episodes actually come from, the two groups were no different, and a greater proportion of the sounds produced a response in the sleepwalkers. They hear you. They respond. The response is just not reliably the one you were hoping for.

04When waking is the safer call

Here is where the standard advice runs out, and it runs out in exactly the situation that made you look it up.

Guiding is the default because it assumes something: that you have a bed to steer them toward and enough seconds to get there. When that assumption fails, the conservative option flips, and it flips completely. Waking somebody costs a confused, unhappy few minutes. The alternatives in the situations below cost considerably more than that.

So, these are the moments to stop steering and start speaking loudly:

They have reached an outside door, a window, a balcony rail or the top of a stairwell, and you cannot get to a position beside them without standing in the way. Distance closes fast in a hallway, and a person with a hand on an exterior door handle is already past the point where a gentle redirect is the cautious choice.

They are on stairs, or about to be. Guiding somebody down a staircase they cannot see is a fall risk for both of you, and a fall is the injury the whole exercise exists to prevent.

They have car keys, or are moving toward a vehicle. Mayo lists driving among the things people do during episodes, and there is no version of steering that manages this one.

They are at the stove, or holding something hot, sharp or heavy.

The redirect is already escalating. If they are pushing back against the hand on their shoulder, getting louder, or turning toward you, that is a signal to stop steering rather than to steer harder. Both MedlinePlus and Mayo list becoming aggressive or violent while briefly confused as a real feature of these episodes, and pressure applied to somebody in that state is the thing that provokes it.

Somebody else in the house is about to get hurt, especially a smaller child.

When you have decided to wake them, do it from a step or two away and slightly to the side rather than face to face. Voice first, at normal speaking volume or louder, using their name and then your own name, before any hand reaches them. Say who you are more than once. That is not a formality, and the next section is about why. If you do need to make contact, one hand on the shoulder from beside them beats a grip on the wrist or a hold from the front, which reads as being seized. Then expect the aftermath the NHS describes: fright, confusion, possibly some anger, and a person who has no idea why they are standing in the kitchen. That is the normal, harmless version of this, and it passes.

One boundary worth drawing once, because the two get confused constantly: somebody physically performing a vivid dream in the last third of the night, arms swinging at a threat they can describe when they surface, is in a different situation from the one on this page, and our guide to acting out dreams while asleep covers it properly.

05Why they may not know who you are

If you have ever been shoved, sworn at or swung at by somebody who does not do those things, there is an explanation, and it is a genuinely reassuring one.

Start with what does get through. Idir and colleagues published three studies in Sleep in 2025 on exactly this question, whether you can talk to somebody during one of these episodes. Sixty one adults with the condition were asked. Eighty one percent said they had held conversations during episodes at least occasionally, and talking to a partner turned out to be the third most common behavior of an episode, after talking to nobody and screaming. The picture of a sealed off person who cannot be reached is simply not what happens.

Of the forty eight who answered the detailed questions, half said the words they heard always matched what had actually been said, and another third said they usually did. Your sentence arrives.

The person saying it is the part that goes missing. Asked whether the identity of the person they were talking to matched reality, a quarter said always and another quarter said usually. The remaining half said it was most often, or always, somebody else entirely. And their replies came from somewhere other than the room: only two of the forty eight said their answer was always fitted to the real world, while forty six percent said it mostly came from a dream and twenty seven percent said it always did. The authors conclude that these episodes are not one uniform state at all, but a range of states of consciousness with different levels of responsiveness and a tangled interplay between what is inside and what is outside.

Those are patients referred to specialist sleep clinics in France, the severe end of this, and the first study rests on what people remembered afterward, so take the shape rather than the decimal places. The shape is this: your voice is likely to reach them, your identity is not, and their reply is coming from a scene you cannot see.

What is in that scene matters, because it is not neutral. Castelnovo and colleagues interviewed forty five patients in 2021 and found a sharp split by age. More than two thirds of the children could recall no mental activity from their episodes at all. More than two thirds of the adults recalled at least one. Seven adults and one child described vivid hallucinatory experiences of dreamed objects or characters projected onto their real home environment, with no reality testing at all. The content was dominated by actions performed from the person's own point of view, often with apprehension, in response to misfortune and danger, in a home setting.

Earlier work says the same thing in more detail. Oudiette and colleagues, writing in Sleep in 2009, collected one hundred and six such reports from forty three adults with severe sleepwalking or sleep terrors. Seventy one percent of the patients reported at least one, ninety five percent of the reports were a single visual scene rather than a story, and the emotional weather was consistent: apprehension in eighty four percent, misfortune in fifty four percent, aggression in twenty four percent. In every single case involving aggression, the sleeper was the one being attacked. The scenes they described are a ceiling coming down, a truck about to run them over, a baby who has to be carried out of the room.

Put those two findings side by side and the hallway makes sense. An adult mid episode may be seeing something frightening laid over the actual room, and they are the one in danger in it. A hand arriving out of the dark gets absorbed into whatever is already happening. So, if you were hit, you were not hit by somebody expressing anything about you. You were, at worst, part of the scenery in a bad scene. That is worth saying out loud to a partner who has been quietly turning it over.

This is also, incidentally, why sleepwalking is not simply unconsciousness. Idir, Oudiette and Arnulf described the state in a 2022 review as a dissociated one, with wake like activity in the motor and emotional regions of the brain and sleep intensity preserved, or even increased, across the networks that handle judgment and self awareness. Awake enough to walk, open a door and answer you. Asleep enough that none of it is being checked against the world, or written down. A different boundary state, sleep paralysis, is the mirror image of it, and we cover that in our guide to sleep paralysis.

06How long the confusion lasts, and who is actually rattled

Short, in both directions. The episode itself is usually under ten minutes. The disorientation afterward, whether it ends on its own or because you woke them, is a matter of minutes rather than hours, and there is no research showing a lasting effect from either.

They will very probably remember none of it. The NHS says flatly that you do not remember sleepwalking when you wake up, and partial or complete amnesia for the episode is one of the formal diagnostic features of this family of conditions. Which means the person who has just had the frightening experience is you.

That asymmetry is worth naming, because nobody writes guidance for the observer. Watching somebody familiar move through the house like a stranger is unsettling in a way that is difficult to explain the next morning, and there is no need to be brisk about it. You are allowed to sit down for ten minutes afterward.

When you do tell them in the morning, tell them plainly and without the teasing version. Mayo lists being very embarrassed about their actions among the real consequences of this condition, and if the person who sleepwalks is the one reading this: you did not choose any of it, there is no evidence that what you did or said reveals anything true about you, and shame is doing no useful work here. It is a mechanical problem with how deeply and how cleanly you come out of one stage of sleep. Reading something into the content of an episode is a different exercise from reading a dream, which is ground our page on what nightmares mean already covers.

07Preventing the next one is mostly done before bed

The most useful twenty minutes in all of this are not spent in a hallway at midnight. They are spent on an ordinary evening, when nothing is happening.

The NHS list is short and it is the right list. Keep doors and windows closed, and use stair gates for a child who sleepwalks. Keep floors clear of anything somebody could trip over. Lock away sharp objects such as knives and tools. Fit a door alarm if there is any risk of somebody leaving the house. Do not let a child who sleepwalks take the top bunk.

Notice what that does to the earlier decision. Every hard call in the hallway is a call about time. A bolted outside door and a stair gate hand you thirty seconds you did not have, and thirty seconds is the difference between a situation that demands you wake somebody and a situation where you can simply take an elbow and walk them back to bed. The guide or wake question gets very easy once the house is doing half the work.

08What makes an episode more likely

Sleep debt is the best supported answer, by a distance. Zadra and colleagues took forty adult sleepwalkers into a laboratory with video and recorded them on a normal night, then again during recovery sleep after twenty five hours awake. On the baseline night they captured thirty two episodes from twenty of the forty. After the sleep deprivation, ninety two episodes from thirty six of them. Same people, same room, one variable. That is why sleep loss is now used deliberately as a way of provoking episodes for diagnosis, and it is the most actionable single fact here.

Then there is a finding that changes how the question feels. Lopez and colleagues studied a hundred adult sleepwalkers at a French sleep clinic in 2013 and asked what set episodes off. Just under sixty percent could name something, and the paper lists them in order of how often they came up: psychological stress, strong positive emotions, and sleep deprivation, then less frequently alcohol or hard evening exercise. Strong positive emotions outranks sleep deprivation in that sentence, and the example the authors give is an unusually exciting leisure activity.

That means a wedding can do it. A birthday, a school trip, a first day somewhere new, a very good day. Nothing has to be wrong for an episode to happen, which is easy to lose sight of when you are lying awake afterward auditing the week for what went wrong. Sometimes the honest answer is that something went right and the night could not settle.

The rest of the list is familiar: not enough sleep, illness with a high temperature, stress, alcohol, an irregular schedule or travel, and some medicines including sleeping pills. Sleep apnea and restless legs appear because both fragment the deep sleep these episodes come out of, and it runs in families strongly enough that Mayo names that a risk factor in its own right.

As for how common all this is, a 2016 systematic review and meta analysis by Stallman and Kohler in PLOS ONE pooled fifty one studies covering just over a hundred thousand people and estimated lifetime prevalence at 6.9 percent, with a confidence interval running from 4.6 to 10.3. Within the previous twelve months it was 5.0 percent of children and 1.5 percent of adults. The authors flag substantial risk of bias across the underlying studies, so hold those loosely, but the shape is right and it is the shape most parents need: this is a common childhood thing, it is several times rarer in adults, and children usually grow out of it by the teenage years without any treatment at all.

09When it is worth a conversation with a doctor

Not tonight, for most people, and not as an emergency.

The NHS threshold is a practical one. See a doctor if the sleepwalking is disrupting sleep, if you are worried about accidents or injuries, or if it comes with other symptoms such as heavy daytime tiredness or breathing that stops and starts during sleep. Mayo adds a few more: episodes more than once or twice a week or several times a night, behavior that has led to danger or injury, sleepwalking continuing into the teenage years, or sleepwalking starting for the first time in adulthood.

That last one sounds ominous and reads worse than it is. Mayo's own phrasing is that when it starts in an adult it is more likely to be related to other underlying conditions, which is a reason to make an appointment rather than a verdict. In practice it often means the appointment finds something treatable, and sleep disordered breathing is high on that list.

Medicines deserve their own line. Some sleeping pills are associated with this, and if episodes started around a new prescription or a dose change, that is genuinely useful information. The instruction from every source, including the NHS, is the same: take the dates to the prescriber and do not stop or adjust anything on your own to test the theory.

It also helps to know who ends up in a sleep clinic, so you can place yourself outside that group. Lopez and colleagues' hundred patients were the severe end by design, referred because episodes were frequent, dangerous or disturbing. Fifty eight percent had a history of violent or dangerous behavior in sleep, ten of them had been hurt badly enough at least once to need medical care, and in six cases a bed partner needed medical care. The paper says outright that people who consult a sleep laboratory may not represent sleepwalkers in general, and that is the important sentence. Those numbers describe a referred clinic population, not the seven year old in your hallway. What they do establish is that this is a real condition with real treatment at the severe end, which is an argument for the appointment rather than for worry.

If you go, bring dates. Time of night, what they did, whether they left the room or the house, anything that nearly went wrong, how much sleep they had had, illness, alcohol, every medicine and when it is taken, and whether they remembered anything at all. Mainieri and colleagues encourage overnight video sleep studies and home video recordings in the diagnostic work up, so a clip a partner already has on their phone is genuinely useful history, taken with the sleeper's agreement, since they cannot give it in the moment. Two weeks of dated notes tells a clinician more than the best retelling of one alarming night.

And the obvious exception: if somebody is hurt tonight, or has got out of the house, that is a now problem and not a wait and see one.

10Common questions about waking a sleepwalker

Is it dangerous to wake a sleepwalker?

No. MedlinePlus states directly that it is not dangerous to awaken a sleepwalker, and the only consistently reported consequence is confusion or disorientation for a short time. The NHS adds that somebody woken suddenly may be scared, angry or upset. That is unpleasant for a few minutes and it is not harm.

If you wake up a sleepwalker, can they die?

No. There is no described mechanism by which being woken kills or damages somebody, and no clinical source reports one. The deaths and serious injuries connected with sleepwalking come from falls, stairs, windows and leaving the house during an episode, which is an argument for securing the house rather than for standing still in a doorway.

Should I guide them back to bed, or just leave them alone?

Guide them if there is anything nearby that could hurt them, and leave them if there is not. Undisturbed episodes usually end with the person going back to sleep on their own, though possibly not in their own bed. Guiding is only worth doing when it improves on that, which mostly means moving them away from stairs, doors and hard edges.

Can a sleepwalker hurt me if I wake them?

It happens, and it is more likely if you startle them, block them or hold them. MedlinePlus lists aggressive behavior when woken by somebody else among the features of sleepwalking, and Mayo notes that people can become violent while briefly confused. Speaking from a short distance before making contact, and approaching from the side rather than the front, is how you reduce that.

Will they remember being woken?

Usually not, or only in fragments. Amnesia for the episode is one of the defining features of these conditions, and it applies whether the episode ended on its own or because you intervened. If you need them to know what happened, tell them in the morning.

Can I wake my child while they are sleepwalking?

Yes, and the same logic applies as with an adult, which is that it is safe but usually unnecessary. Guide them back to bed instead if you can. Children are also the group least likely to have any mental content during an episode, so there is generally less to frighten them out of than there would be with an adult.

What do I do if they are trying to leave the house?

Wake them, and start with your voice from a step or two away rather than a hand from behind. This is the clearest situation where the default flips, because outside is where the serious injuries happen and because a door closes the distance faster than a redirect can work. Then fit a door alarm or a bolt at an unfamiliar height before the next time.

Are they dreaming?

Partly, and this is a more interesting answer than it used to be. Sleepwalking comes out of deep non REM sleep rather than the vivid dreaming stage, so it is not a dream being performed. But it is not a blank either. Oudiette and colleagues found that seventy one percent of adult patients could report a short, usually frightening scene attached to their episodes, and later work has found adults describing dreamed figures projected onto their actual bedroom. Children mostly report nothing at all.

If you came here mid episode, the useful part is short: steer them if you can, wake them if you must, and check the doors and the stairs before you go back to bed. If you came here the morning after, still turning over what they said or what they did, the useful part is that none of it was aimed at you and none of it means anything about them. DreamTold exists for the part of the night that is worth thinking about, so once the safety questions are settled you can go back to the dreams themselves in the dream dictionary. This page is educational and written for reflection rather than as medical advice, and it cannot diagnose anybody. If episodes are frequent, if anyone has been hurt, if this began in adulthood, or if it started alongside a new medicine, that is a conversation for a qualified clinician.

11Sources

Every claim here that comes from somewhere else, with the somewhere else, so you can read the original.

• NHS: Sleepwalking, including what to do when someone is sleepwalking
• MedlinePlus: Sleepwalking
• Mayo Clinic: Sleepwalking, symptoms and causes
• Neurology: Pilon and colleagues, precipitating factors of somnambulism, the impact of sleep deprivation and forced arousals, 2008
• Sleep Medicine: Pilon and colleagues, auditory arousal responses and thresholds during REM and NREM sleep of sleepwalkers and controls, 2012
• Annals of Neurology: Zadra and colleagues, polysomnographic diagnosis of sleepwalking and the effects of sleep deprivation, 2008
• Sleep: Idir and colleagues, talking to sleepwalkers, response to communication efforts in disorders of arousal, 2025
• Sleep: Oudiette and colleagues, dreamlike mentations during sleepwalking and sleep terrors in adults, 2009
• Nature and Science of Sleep: Castelnovo and colleagues, mental activity during episodes of sleepwalking, night terrors or confusional arousals, 2021
• Journal of Sleep Research: Idir, Oudiette and Arnulf, sleepwalking, sleep terrors, sexsomnia and other disorders of arousal, 2022
• Sleep: Lopez and colleagues, functional impairment in adult sleepwalkers, a case control study, 2013
• PLOS ONE: Stallman and Kohler, prevalence of sleepwalking, a systematic review and meta analysis, 2016
• Diagnostics: Mainieri and colleagues, diagnosis and management of NREM sleep parasomnias in children and adults, 2023