Falling asleep is a handover rather than a switch, so your body has to lower its guard, and a stress response is built to keep that guard up instead.

Researchers split the pre-sleep state into two kinds of arousal, one of them thinking, the busy commentary in your head, and one of them physical, the warm chest.

In the lab, people with long-running sleep trouble show more of the main stress hormone than good sleepers, and the gap is widest in the evening and the first half of the night.

That is the opposite of the popular story, because the night-time version of stress is a system running too high at the hour you most need it to come down.

The two kinds of arousal feed each other, because a racing body gives your mind something to explain, and the worry then keeps the body switched on.

Waking in the night is part of the same picture. In one study, the evening stress hormone level tracked the number of times people woke, in good sleepers as well as poor ones.

Timing matters more than most people expect, because worry arriving close to sleep does more damage to the night than the same worry earlier in the evening.

This is the loop worth breaking, and it is why lowering the whole state beats arguing with each thought as it turns up. You are not talking yourself down from one worry, you are lowering the setting that made it loud.

Sleep is not a switch you flip, it is a handover, and it only happens once your body agrees that nothing needs watching. Temperature drops, the pulse slows, and attention loosens its grip on the outside world.

A stress response is built to do the exact opposite, raising alertness and holding it there, because the job it evolved for was short and urgent. It was never designed to run through a whole evening.

So the two systems are asked to work at once. You are lying still in a dark room while the part of you that handles threat is still on shift. Nothing is broken, because two important jobs are simply in conflict with each other.

Researchers describe the pre-sleep state as two separate things. One is the thinking kind, the running commentary and the list. The other is the body kind, the warm chest, the tight stomach, the pulse you can suddenly hear. Most people have both, and they are measured separately for good reason.

There are two routes out of the brain when a threat is registered, and the nerve route is fast and lifts your pulse and muscle tone within seconds. The hormone route is slower, and it keeps the setting raised for a while afterwards.

The clearest sign that thought alone can do this comes from a small lab study. Fifteen healthy adults slept in a laboratory and were simply told that they would be woken early, and nothing else about the night was changed at all.

On those nights the main stress hormone began climbing about an hour before the early wake time, and it did not climb on nights when they expected to sleep on. An expectation about tomorrow, held in mind, moved a hormone overnight.

Later work pushed further. When people knew a demanding speaking task was waiting for them after a nap, their sleep was measurably shallower, and how close the task was mattered. Fifteen and sixty-four people are thin samples, so hold the detail loosely and the direction firmly.

Plenty of people rule themselves out here, because they are not panicking and do not feel especially anxious, and they are simply awake and thinking about something ordinary at an unhelpful hour.

That mild version is the common one, and it happens because the amount of stress needed to change a night is small. In a study of nearly 2,900 people who slept well at the start, the chance of new sleep trouble a year later rose by around a fifth for each extra stressful event they had faced.

The same work found that how long a stressor dragged on mattered as much as how big it was. A slow, unresolved worry did more than a sharp one that ended.

The other clue is that it is not the count of problems. Over three weeks of daily diaries, people who slept badly reported the same number of minor hassles as good sleepers. They rated the impact higher, and had more going on in their heads at bedtime.

Start with how common the waking is. In a representative sample of nearly nine thousand US adults, more than a third woke at least three nights a week, and most said it had been happening for over six months. so the waking itself is not the unusual part of this.

Then the everyday inputs, which get blamed far less than they deserve, starting with caffeine, which still disturbed sleep in a trial when taken six hours before bed. A single drink settles the first half of the night and then disrupts the second half of it.

Bright screens in the last hour push your internal clock later, so you are lying there earlier than your body thinks it is. A full bladder, a warm room, hot flushes and a new medication all wake people too.

None of that is a reason to stop looking at stress. It is a reason to stop treating stress as the only suspect. A hard week with late screens, more coffee and a glass of wine is four causes arriving together, not one.

The part that turns a bad fortnight into a bad year is learning. Do enough tense nights in the same bed and the room itself starts to bring the alertness with it. The particular cue and the response gradually become wired together.

This is also where the good news lives, because a learned pattern can be unlearned. Structured behavioural treatment for sleep is the best-tested route, and it works on the worry as well as the sleep.

Pooling fifty studies and 2,690 people, that treatment produced a small to moderate improvement in the anxiety that came alongside the sleep problem. The reviewers were careful to say weaker studies may have flattered the result.

There is also a trait side to this. How much your sleep reacts to stress runs in families, is partly inherited, and can be measured before anything goes wrong. If yours is reactive, that is worth knowing, and it is a reason to protect the wind-down rather than proof that you are stuck with it.

One popular explanation deserves retiring. The 3am wake-up is often sold as burnt-out adrenal glands, or a hormone tank that has run empty. A systematic review that screened thousands of papers and kept 58 found no substantiation for that idea at all.

The measured picture is close to the reverse, because the stress system in poor sleepers is running high in the evening and the first half of the night rather than running out, and empty and overactive lead to very different advice.

The evidence also argues with itself in a place that matters. In one lab study of forty good sleepers, being told about a demanding task the next day changed their morning hormone response but left their sleep untouched, by diary and by monitor.

And in a fourteen-day trial of 186 people, moving worry to a set slot earlier in the evening cut how long people worried, yet did not improve any sleep measure. So the mechanism is solid, and the promise that managing worry fixes the night is not. That is the honest state of it.