Most nightmares come out of dreaming sleep, which builds up in the second half of the night. That is why they land at 4am rather than at midnight, and why a long lie-in can hand you the worst one.

In dreaming sleep the brain runs emotional material with its alarm chemistry turned down. Researchers describe that as a nightly window for taking the sting out of what happened to you.

A stressed day gives that window more to carry. One well-known model calls this the load side of nightmares: more emotional pressure in the day, more to process at night.

The other half is how strongly a person reacts. Two people can carry the same day and only one of them wakes up with the feeling still running.

Stress also changes the setting. In sleep-lab work, people with frequent nightmares had more small arousals in the ten minutes before dreaming sleep began.

That matters because broken dreaming sleep does the processing job badly. The material gets stirred rather than settled, and it comes back the next night.

Then the loop closes from the other side. In a diary study of nurses, a night with a nightmare was followed by a more stressful day, which fed the next night.

So the useful target is not the dream. It is the load going in and the state of the sleep it lands in, and both of those you can actually move.

Sleep runs in cycles of about 90 minutes, and the dreaming part of each cycle gets longer as the night goes on. The bulk of it sits in the last few hours before you wake, which is where nightmares mostly live.

That stage is built for emotional housekeeping. Body movement is switched off, the alarm chemistry that keeps you vigilant while awake runs low, and the brain replays material from waking life in a setting where nothing can happen to you.

Because of the timing, small things shift the odds. A late lie-in gives you more of that stage. So does catching up after short nights, which is why the bad dream often arrives on the recovery night rather than the exhausted one.

In a 16-country survey of 15,854 adults, the link with sleep length was a curve rather than a line: people sleeping very little and people sleeping a lot both reported more nightmares than people in the middle. The middle is roughly the seven to nine hours most adults are aimed at.

Two things travel from the day into the night. The fast one is your state at lights out, still alert, still braced. The slow one is the pile of feeling the day left unfinished, which dreaming sleep works through hours later.

You can show the second one in a lab. In 2025, researchers put 55 people through either a psychological stress task or a relaxation task, then woke them repeatedly and collected 668 dream reports.

The timing moved with the stressor. Stress before sleep made early dreams harsher, while being told a stressful task was waiting in the morning made late dreams harsher. Nothing physical was done to anyone: the whole trigger was mental.

Older work points the same way. Watching a distressing film before bed raised dream anxiety compared with a neutral film, and in a trial of 146 people, five minutes of chewing over an intrusive thought at lights out raised the chance of dreaming about something threatening.

People often exclude themselves here, because they picture a screaming nightmare and theirs are just long, unpleasant, exhausting dreams. The research does not draw that line as sharply as the word does.

In diary studies the graded version is what shows up. Higher-stress days raised both the odds of a nightmare and how bad it was rated, so this behaves like a dial rather than a switch.

In one two-week diary study of 392 nurses, nearly half reported at least one nightmare in the fortnight. Those are working adults, not a clinic sample, and the pattern held whether they had trauma symptoms or not.

The cost is also not really in the dream. It is the waking up, the shortened sleep that follows, and the next day being harder than it needed to be.

Stress is one input among several, and the others are worth naming because they are adjustable. Sleep length is the first: both very short and very long nights sit higher for nightmares.

Alcohol is the most reliable everyday one. A drink suppresses dreaming sleep early in the night, then the body takes it back in the second half, and that rebound is when the vivid, unpleasant dreams show up.

Medicines matter too. In a review of nearly 127,000 safety reports for one family of blood-pressure drugs, about 1 in 100 described nightmares, more often with the fat-soluble ones that reach the brain easily.

Then breathing and digestion. Snoring and pauses in breathing travel with nightmare frequency in community surveys, and eating close to bedtime leaves the gut working while you sleep. None of that makes stress irrelevant, and none of it makes stress the only cause.

Here the news is genuinely good. Nightmares respond to a simple trained skill: while awake, you write the dream out and rewrite the ending you want, then rehearse the new version for a few minutes a day.

Pooled across 15 trials and 1,078 adults, that approach gave a moderate drop in how often nightmares happened, on a par with the leading medicine for them. A separate review of 13 studies found gains that were still there 6 to 12 months later.

It also works without a clinic. In a randomised trial, people worked through the rewriting exercise from a self-help workbook and still improved, which is unusual in sleep research.

The honest caveat: the strongest trials were run in people whose nightmares followed frightening events, and in the one study with an active comparison the advantage was small. Expect weeks of practice, not a single good night.

One popular belief deserves retiring: that a late cheese snack causes nightmares. In a 2025 survey of 1,082 people, only about 1 in 20 thought food changed their dreams at all, and the dairy complaints clustered in people who do not digest it well.

So the route is night-time discomfort and worse sleep, not cheese acting on dreams. The same goes for the tidy idea that a nightmare is a literal replay of what happened to you; most are reworked, not repeated.

The evidence also disagrees with itself in a useful place. In a 30-day diary study of people with regular nightmares, fear of going to sleep tracked their distress, but how wound up they were at lights out did not predict the nightmares at all.

And in that 2025 lab study, stress changed how intense the threat in a dream felt but not how often threatening content appeared. So hold the mechanism firmly and the detail loosely: stress shapes the dream, and exactly which part it shapes is unsettled.