Stress and Deep Sleep: What Studies Show About Worry
You know the kind of night. There's a presentation tomorrow, or a bill you haven't paid, or a conversation that keeps replaying. You do fall asleep in the end, and in the morning your ring or watch shows only a thin sliver of deep sleep. Was it the worry, or just a bad night?
The research is more interesting than the slogan that stress ruins sleep. Some studies find a clear drop in deep sleep and others find nothing. The timing of the stress also seems to matter more than most people assume.
What deep sleep is, briefly
Sleep scientists call deep sleep N3, or slow-wave sleep, after the large, slow brain waves it produces on an EEG. Most of it happens in the first part of the night. In a lab it's measured with polysomnography (PSG), which records brain activity directly. Your wearable can't do that. It infers sleep stages from movement and heart signals. Keep that difference in mind, because nearly all the stress research below used PSG.
The big picture: stress often, but not always, cuts deep sleep
The most cited overview is a systematic review by Kim and Dimsdale in Behavioral Sleep Medicine that gathered 63 polysomnography studies of psychosocial stress. In the experimental studies, stress fairly consistently meant less slow-wave sleep, less REM sleep, lower sleep efficiency and more awakenings.
Look closer, though, and the deep sleep evidence is mixed. Among the experimental studies that reported slow-wave sleep, about a third found a decrease, none found an increase, and roughly two thirds found no change. Many studies had only around 10 participants. The authors said they were surprised how few studies existed. They also pointed out that a stressor only counts if the person actually finds it stressful, and most studies never checked. Two studies in which people expected to give a speech, the closest the review gets to classic bedtime dread, found no change in slow-wave sleep.
So it's fair to say stress tends to lower deep sleep. It just doesn't happen every time, or to everyone.
Timing matters: stress before bed vs. dread about tomorrow
A newer study adds detail. Beck, Loretz and Rasch, in Cerebral Cortex, recorded 67 healthy young women with PSG during 90 minute daytime naps in the lab. Each woman had two sessions. One came after a relaxing virtual nature scene. The other involved a stressful task in virtual reality: a speech to a panel followed by mental arithmetic. Half the women did the stressful task right before the nap, and the other half were told they'd do it after waking up.
Both versions reduced slow-wave sleep. Across both groups it fell from about 17.6 minutes after relaxation to about 13.1 minutes under stress. Falling asleep also took longer, roughly 19 minutes instead of 12.
The groups differed in when the effect appeared. Stress just before sleep raised heart rate and physical arousal, and the drop in deep sleep came early in the nap. Expecting stress after waking raised mental arousal but not heart rate, and its effect came later in the nap. The authors suggest the later effect may come from the brain repeatedly replaying thoughts about the upcoming event during sleep. That's their hypothesis, not something the study proved.
The caveats are real. A nap isn't a full night, everyone in the study was a young woman, and the comparison was relaxation rather than a neutral condition, which the authors say could make the stress effect look bigger than it is.
Outside the lab: bedtime worry predicts a worse night
Labs control everything and real life doesn't. Åkerstedt and colleagues in Sleep Medicine followed 50 people for 42 days with a daily sleep diary and stress questions. Stress and worry at bedtime were the main predictors of poorer sleep quality the next morning, along with factors like how long and how well people had slept the night before. The researchers noted that stress levels in their sample didn't vary much, so the true effect may be larger.
This was self-rated sleep quality, though, not measured deep sleep. It shows that worry and a bad night go together from day to day. It can't tell you how many minutes of N3 you lost.
Can writing it down help?
One small, well-designed experiment tested a cheap idea. Scullin and colleagues in the Journal of Experimental Psychology: General randomly assigned 57 healthy adults aged 18 to 30 to spend five minutes before lights out writing one of two lists. One group wrote a to-do list for the coming days. The other listed tasks they had already finished. Sleep was recorded with PSG.
The to-do group fell asleep faster: about 16 minutes on average, compared with about 25. People who wrote more specific to-do lists fell asleep faster still. Deep sleep didn't change, though. N3 made up about 15 to 16 percent of the night in both groups.
This was a single night in a lab with healthy students, and there was no group that skipped writing. So we can't tell whether to-do lists help or lists of finished tasks hurt. It's still low risk and easy to try.
What your wearable can and can't see
If you want to know whether worry affects your own deep sleep, it helps to know what your device does well. Robbins and colleagues in Sensors compared an Oura Ring Gen3, an Apple Watch Series 8 and a Fitbit Sense 2 with PSG over one night in 35 healthy adults. All three told sleep from wake well, but deep sleep was harder. On average the Apple Watch underestimated deep sleep by about 43 minutes and the Fitbit by about 15, while Oura's average matched PSG. Still, night by night agreement on deep sleep was poor for all three devices. Oura funded the study, the lead author sits on Oura's medical advisory board, and Whoop wasn't tested.
The practical upshot is that one night's deep sleep number is noisy. Your own trend over many nights, on the same device, tells you more than any single reading or a comparison with someone else's watch.
A sensible way to experiment
- Pick one change to your wind-down, like a five minute to-do list or a short breathing exercise, and keep everything else the same.
- Give it at least two weeks, because single nights vary too much to judge.
- Compare against your own normal nights, not population averages.
- Note how you feel too: how long it takes to fall asleep, your mood in the morning, your energy during the day.
- Treat a small change as a maybe until it shows up again.
This kind of test is what Somnia Deep Sleep, an iPhone app now in beta, is being built for. You change one habit for 14 nights, and it compares the deep sleep your Apple Watch, Oura or Whoop already writes to Apple Health against your own 30-night baseline. Each result gets a label: Strong signal, Likely or Too early to tell.
What the research can't tell you
Most studies on stress and deep sleep are small and short. They usually take place in labs, often with young, healthy volunteers, and sometimes with naps instead of full nights. The first night in a sleep lab is mildly stressful in itself, which muddies the results. Each study defines stress in its own way, and few have tracked measured sleep stages at home over weeks.
We also don't know how much a modest drop in deep sleep after one anxious night matters for your health or how you feel the next day. And there's no good evidence that any single bedtime trick reliably raises deep sleep for everyone.
When to talk to a doctor
Lying awake worrying now and then is common. If trouble sleeping has gone on for three months or more, or is making daytime hard to cope with, see a doctor. The NHS notes that you may be offered cognitive behavioural therapy. Loud snoring with gasping or choking, breathing that stops and starts, or daytime sleepiness that affects your driving are also reasons to get checked, as the NHS page on sleep apnoea explains. A wearable can't diagnose any of these.
Sources
- Kim EJ, Dimsdale JE. The effect of psychosocial stress on sleep: a review of polysomnographic evidence. Behavioral Sleep Medicine. 2007;5(4):256-278. PMC4266573
- Beck J, Loretz E, Rasch B. Stress dynamically reduces sleep depth: temporal proximity to the stressor is crucial. Cerebral Cortex. 2023;33(1):96-113. Oxford Academic
- Åkerstedt T, Orsini N, Petersen H, Axelsson J, Lekander M, Kecklund G. Predicting sleep quality from stress and prior sleep: a study of day-to-day covariation across six weeks. Sleep Medicine. 2012;13(6):674-679. Europe PMC record
- Scullin MK, Krueger ML, Ballard HK, Pruett N, Bliwise DL. The effects of bedtime writing on difficulty falling asleep: a polysomnographic study comparing to-do lists and completed activity lists. Journal of Experimental Psychology: General. 2018;147(1):139-146. PMC5758411
- Robbins R, Weaver MD, Sullivan JP, et al. Accuracy of three commercial wearable devices for sleep tracking in healthy adults. Sensors. 2024;24(20):6532. PMC11511193
- NHS. Insomnia. nhs.uk
- NHS. Sleep apnoea. nhs.uk
Wondering whether a bedtime to-do list or a calmer wind-down changes your own deep sleep? Join the Somnia beta and test it on your own nights.
This article is general information, not medical advice.
Test it on your own nights
Somnia runs 14-night habit experiments on the deep-sleep data your Apple Watch, Oura or Whoop already records, then tells you how confident the result is.
Join the beta

