A child sits up ninety minutes after bedtime and screams. The eyes are open and wide, the heart is hammering, the skin is damp, the breathing is fast, and nothing said from the doorway lands. Then, after a couple of minutes, it stops. They lie back down and sleep. In the morning they are perfectly cheerful and have no idea what you are talking about.
That is a night terror, and the most useful thing to say about it is the thing its name gets wrong. It is not a bad dream. A nightmare is a REM event with a story in it, remembered on waking. A night terror is an incomplete arousal out of the deepest stage of NREM sleep: no story, no waking, no memory afterwards, and a body running a full alarm response while the mind that would have to be frightened is still offline. This page covers what is actually happening during an episode, why comforting doesn’t work and sometimes makes it worse, why nearly every child grows out of this, the one behavioural lever with any research behind it, and where the line sits between a household event and a clinical one.
Not a nightmare, and not a milder version of one
The two get filed together because both are frightening to be near. Mechanically they have almost nothing in common, and separating them is the whole reason this page exists.
Nightmares arrive out of REM sleep, which is concentrated in the last hours before morning. They wake the dreamer properly. There is content: a place, a threat, a sequence of events that can be described over breakfast and, if it recurs, worked on directly. The person is oriented within seconds of waking, knows where they are, and can be comforted, because they are awake and comfort is a thing they can perceive.
A night terror inverts every one of those. It comes out of slow-wave sleep in the first third of the night. It does not wake the sleeper. There is no narrative, at most a fragment: a wall coming closer, a shape, a sense of being buried or trapped, nothing with a plot. The sleeper is unresponsive or barely responsive, does not recognise the room or the people in it, and afterwards remembers none of it. Sleep medicine classifies it alongside sleepwalking and confusional arousals as a disorder of arousal, and that grouping is the diagnosis in miniature. These are not dreams being lived out. They are failed transitions.
The practical version of the distinction, for anyone standing in a hallway at eleven at night deciding what they are looking at: check the clock and check the morning. Early plus forgotten is an arousal disorder. Late plus described in detail is a dream.
An alarm with nothing behind it
The terror in a night terror is entirely physiological. Heart rate can more than double. Breathing goes fast and shallow, the pupils dilate, the skin flushes and sweats, and the scream is real. Every part of that is the body’s threat response firing at full strength.
What is missing is the threat, and also the person who would perceive one. During an episode the arousal systems in the brainstem and the circuits that drive movement and autonomic output come back online, while the cortical machinery responsible for orientation, judgement, and laying down memory stays in slow-wave sleep. You are watching an alarm without a subject. This is why the question everyone asks afterwards, what were you dreaming about, so reliably returns nothing. There was no dream to have.
It is also why the episodes cluster where they do. Slow-wave sleep, the deep sleep stage dominated by delta waves, is front-loaded into the first two or three cycles of the night, because sleep pressure is highest right after lights-out and the body spends it early. An arousal out of slow-wave sleep can only happen where slow-wave sleep is. By the second half of the night the sleep cycle has handed most of its time over to REM, and whatever wakes someone screaming at five in the morning is a different phenomenon with a different explanation.
Why comforting doesn’t work
This is the part that costs parents and partners the most, so it is worth being direct about. Holding the person, saying their name, turning on the light, and telling them they are safe does not calm a night terror, because none of it is being received. Attempts at physical comfort are sometimes met with pushing away, thrashing, or an escalation in the screaming, and the person doing the comforting reasonably concludes they have made it worse. Often they have, though not through any failure of technique.
The alternative is unsatisfying and correct: stay, and do very little. Sit or stand close enough to stop a fall or move something hard out of the way. Keep your voice low, or say nothing at all. Don’t restrain, don’t argue, don’t try to wake them unless the alternative is an injury. Most episodes end within a few minutes, and the sleeper lies down and continues the night without a break in it.
Waking someone mid-episode is not dangerous, and the folklore that says otherwise is wrong here for the same reason it is wrong about sleepwalking. But it is usually the worse option, because a person pulled out of a deep arousal is disoriented for a stretch that can run considerably longer than the episode would have. You trade two minutes of noise for twenty minutes of confusion.
Who gets them, and what raises the odds
Night terrors are largely a childhood phenomenon, most common in the preschool and early school years, and the great majority of children stop having them by adolescence with no intervention at all. Adults have them far less often, for the straightforward reason that adults have less slow-wave sleep to be dragged out of.
There is a strong familial pattern, as there is across the arousal disorders. Households that produce night terrors often produce sleepwalkers and confusional arousals in the same generation or the one before it. What appears to be inherited is the tendency toward unstable exits from deep sleep, not the specific behaviour, which is why a parent who sleepwalked as a child may have a child who screams instead.
Sitting on top of that predisposition is a short list of things that make an episode more likely on any given night.
Sleep debt is the most reliable of them. After short nights the body pushes slow-wave sleep deeper to compensate, and a deeper stage is harder to exit cleanly. This is the loop worth understanding, because it runs both ways: a disrupted night raises the odds of the next episode, and the episode disrupts the next night.
Irregular timing does similar damage without the shortfall. A bedtime that swings by two hours between weeknights and weekends leaves the early cycles arriving at unpredictable points, and the transitions ragged.
Fever and illness are a classic childhood trigger and almost always self-limiting. A run of episodes during a virus is rarely the beginning of anything.
Anything that nudges the sleeper supplies the arousal an episode needs: a full bladder, a noisy street, a sibling, a pet on the bed, a room that gets too warm. In adults, untreated sleep apnea is the version of this that matters most, and treating the breathing sometimes ends the episodes outright.
Some medications, particularly sedatives and certain psychiatric drugs, can bring on complex sleep behaviours including terrors. If episodes started after a new prescription, that belongs in a conversation with the prescriber rather than in a bedtime routine.
Stress raises frequency, and this is where the genre tends to overreach. Elevated stress makes episodes more likely; it does not make them meaningful. A night terror is not the subconscious surfacing something. Unlike dream content, which at least has material to interpret, an arousal disorder has nothing in it to read.
Scheduled awakenings, honestly described
There is one behavioural technique with actual support behind it, and it works by pre-empting the arousal rather than treating anything.
For episodes that arrive at a fairly consistent clock time, the approach is to wake the sleeper briefly, around fifteen to thirty minutes before the usual time, every night for two to four weeks. Not a full waking: enough to stir them, a hand on the shoulder and a quiet word, then let them settle. The idea is to interrupt the build toward the arousal and shift the architecture of that part of the night just enough that the episode has nowhere to form.
The evidence is real but modest, and it is worth saying which kind. This comes from small studies and case series rather than large controlled trials, mostly in children, and the effect is usually described as episodes becoming less frequent rather than stopping outright. It also depends entirely on the timing being predictable, which is why the log described at the end of this page matters more than it sounds like it should. If episodes land anywhere between ten and two, there is nothing to schedule against.
Everything else on the list is unglamorous and largely borrowed from sleep hygiene: protect the total amount of sleep, hold a consistent wake time, keep the room cool and dark, empty the bladder before bed, and make the physical space around the bed boring enough that thrashing has no consequences.
When it belongs with a clinician
Most night terrors need nothing more than time and a steadier schedule. Some need more than that, and the markers are specific.
Episodes beginning for the first time in adulthood, because new-onset terrors in adults are more often driven by an untreated sleep disorder, a medication, or alcohol than by anything spontaneous. Any episode that has caused an injury, or that involves leaving the bed and moving through the house, which puts it in overlapping territory with sleepwalking. Loud snoring, witnessed pauses in breathing, or heavy daytime sleepiness alongside the episodes, which points at apnea as the arousal trigger and is treatable. Episodes that are very frequent, several in a single night, brief, and near-identical each time, because that stereotyped pattern can indicate nocturnal seizures rather than a parasomnia and needs proper evaluation. And any case where the fear of the episodes has begun to shape the household’s sleep, which is a real problem even when the terrors themselves are benign.
A sleep clinician has tools here that a bedtime routine does not, including a study that can catch the arousals as they happen. The honest position is that some of this is out of reach of anything you can do at home.
How Murmora fits, and where it doesn’t
Murmora does not treat night terrors, and no audio session will stop an episode in progress or shorten one. Anyone claiming otherwise is selling something. What a nightly session can reach is the thing that grows around the episodes: the parent who has started sleeping with one ear open, the partner who braces at the ninety-minute mark, the adult whose own approach to bed now carries a low background dread. That vigilance is expensive, and lost sleep is the factor most consistently linked to more episodes, which makes the watchfulness part of the loop it is trying to guard against.
That is the modest, specific claim worth making. Murmora’s sessions are built around a slow settle at the threshold and a small number of concrete statements, delivered in a guide voice or eventually in a clone of your own, arriving in the minutes when the critical mind has already loosened. For someone whose nights have turned into a shift, a personalized session shaped around I am allowed to rest even when I am on call does more than a generic relaxation track, because the sentence is aimed at the actual problem. The rest of the work still belongs to the earlier bedtime and, where it applies, the clinician.
What to do this week
Start the log, because it is the input everything else depends on. Date, clock time the episode started, roughly how much sleep the night before, and anything unusual: illness, a late night, a new medication, a room that was too warm. Two weeks is enough to see whether the timing is consistent, and timing is both the first thing a clinician will ask for and the thing memory reconstructs worst.
While the log runs, attack the sleep debt, which is the trigger with the most give in it. Move bedtime earlier by half an hour rather than attempting anything clever, and hold the same wake time seven days a week so the early cycles stop arriving at unpredictable points. Do one pass of the space around the bed for anything hard or breakable within reach.
Then, if the timing turns out to be predictable within about half an hour, try scheduled awakenings for a fortnight and keep logging. And in the morning, whatever happened, mention it plainly or not at all. There is nothing to debrief, the sleeper has no memory to consult, and for children the household’s alarm is almost always more distressing than the episode they slept through.