19 cohort studies with 638,792 adults in total, 90,507 of them with insomnia. The authors searched six databases up to August 2024, English-language papers only. Each study grouped people by activity level with its own cut-offs, mostly from questionnaires. Insomnia was defined in different ways: a doctor's finding, a sleep questionnaire, or self-reported complaints. The studies' quality averaged 5.3 on a 9-point scale, which the authors call medium. Medium here means the middle of the low, medium and high levels the paper compares. It describes how active people were, not their sleep.

Compared with low activity: Medium activity: 7% lower odds of insomnia (odds ratio 0.93, 95% range 0.87 to 1.00). The top of the range touches 1, so it's borderline. High activity: no statistically significant link (odds ratio 0.96, range 0.85 to 1.09). High activity didn't do better than medium either (odds ratio 1.03, range 0.97 to 1.09). For total activity only, high went with 5% higher odds than medium (1.05, range 1.01 to 1.09). The authors say the main link was small and close to no effect.

Cause. These are observational studies. They show what went together, not what caused what. Direction. People with insomnia may be too tired to be active, rather than activity affecting sleep. Other habits. People with medium activity may also eat better and have less stress. Mood, stress, income, illness and shift work weren't adjusted for in every study. Consistency. Results varied a lot between studies (I² of 74% in the main analysis). Measurement. Activity was mostly self-reported, and insomnia was defined in several ways. Multiple tests. The authors say the results are exploratory.

Physical Activity and Insomnia: 19 Studies

Q: Does the kind of activity matter?

The paper split high vs low activity by where it happened: At work: 21% higher odds of insomnia (odds ratio 1.21, range 1.02 to 1.43). This rests on 3 studies, and the authors call it hypothesis-generating. Leisure: 15% lower odds, but not statistically significant (0.85, range 0.71 to 1.02). Total activity: no link (0.99, range 0.90 to 1.09). For leisure activity at the medium level, the odds were 11% lower than low (0.89, range 0.79 to 0.99). The authors suggest physical jobs often come with work stress, which leisure activity doesn't. The study didn't test that.

A 2026 meta-analysis of 19 cohort studies with 638,792 adults linked medium activity with 7% lower odds of insomnia than low activity. High activity showed no clear link, and activity at work went with 21% higher odds.

PulseOxTracker Team16 October 2026

Infographic summarising a 2026 meta-analysis of physical activity and insomnia. 7% lower odds of insomnia with medium activity than with low activity. Borderline: the top of the range reached 1.00. Who was studied: 19 cohort studies with 638,792 adults, of whom 90,507 had insomnia; 27 estimates were pooled; activity was mostly self-reported on questionnaires. Odds of insomnia, each with its 95% range. By level: medium vs low, odds 0.93, 7% lower odds; high vs low, odds 0.96, no clear link; high vs medium, odds 1.03, no clear link. By kind of activity, high vs low: activity at work, odds 1.21, 21% higher odds, from 3 studies; leisure activity, odds 0.85, no clear link; total activity, odds 0.99, no clear link. Left of the dashed line means lower odds, right means higher odds. A range that crosses 1 is not statistically significant. Men and women showed no significant difference for high vs low activity. What it cannot show: these are links, not proof of cause. Poor sleep may lead to less activity, rather than the reverse. Results varied a lot between studies. Activity and insomnia were measured in different ways. Mood, stress and shift work were not always accounted for. — A summary of the 2026 meta-analysis: medium activity went with slightly lower odds of insomnia than low activity, high activity showed no clear link, and activity at work went with higher odds.

Exercise is often suggested for people who sleep badly, and trials of exercise programmes back that up. But trials are often small and tightly controlled, so they don't tell you much about everyday activity.

A 2026 meta-analysis looked at everyday life instead. It pooled 19 studies that compared how active people were with whether they had insomnia.

Medium activity went with slightly lower odds of insomnia. More than that showed no clear extra link, and activity at work went the other way.

Who was studied?

19 cohort studies with 638,792 adults in total, 90,507 of them with insomnia.
The authors searched six databases up to August 2024, English-language papers only.
Each study grouped people by activity level with its own cut-offs, mostly from questionnaires.
Insomnia was defined in different ways: a doctor's finding, a sleep questionnaire, or self-reported complaints.
The studies' quality averaged 5.3 on a 9-point scale, which the authors call medium.

Medium here means the middle of the low, medium and high levels the paper compares. It describes how active people were, not their sleep.

What did it find?

Compared with low activity:

Medium activity: 7% lower odds of insomnia (odds ratio 0.93, 95% range 0.87 to 1.00). The top of the range touches 1, so it's borderline.
High activity: no statistically significant link (odds ratio 0.96, range 0.85 to 1.09).

High activity didn't do better than medium either (odds ratio 1.03, range 0.97 to 1.09). For total activity only, high went with 5% higher odds than medium (1.05, range 1.01 to 1.09).

The authors say the main link was small and close to no effect.

Does the kind of activity matter?

The paper split high vs low activity by where it happened:

At work: 21% higher odds of insomnia (odds ratio 1.21, range 1.02 to 1.43). This rests on 3 studies, and the authors call it hypothesis-generating.
Leisure: 15% lower odds, but not statistically significant (0.85, range 0.71 to 1.02).
Total activity: no link (0.99, range 0.90 to 1.09).

For leisure activity at the medium level, the odds were 11% lower than low (0.89, range 0.79 to 0.99).

The authors suggest physical jobs often come with work stress, which leisure activity doesn't. The study didn't test that.

Was it different for men and women?

Not for high vs low activity. The odds ratio was 0.94 in men and 1.03 in women, and neither was statistically significant.

For medium vs low, women had 4% lower odds (0.96, range 0.93 to 0.99). Men's estimate was 0.80, but its range was wide (0.56 to 1.15).

What can't it show?

Cause. These are observational studies. They show what went together, not what caused what.
Direction. People with insomnia may be too tired to be active, rather than activity affecting sleep.
Other habits. People with medium activity may also eat better and have less stress. Mood, stress, income, illness and shift work weren't adjusted for in every study.
Consistency. Results varied a lot between studies (I² of 74% in the main analysis).
Measurement. Activity was mostly self-reported, and insomnia was defined in several ways.
Multiple tests. The authors say the results are exploratory.

The funnel plot looked symmetrical, so there was no sign of missing small studies, but with 19 studies that check is weak. The study was funded by Chinese government science and traditional Chinese medicine programmes, and the authors report no conflicts of interest.

What can you do with this?

The study doesn't show that changing how active you are will change your sleep. You can still watch it for yourself as a self-experiment.

Log your active days and your rest days, and compare how you slept. If you wear an overnight oximeter, the session viewer shows your pulse and oxygen through each night. For a related study, see sitting time and sleep.

If you're thinking about a sleep tracker for insomnia, read the meta-analysis of wearables for insomnia. If poor sleep keeps going for weeks, talk to a doctor.

Not a medical device. Does not diagnose, treat or prevent any condition. Discuss results and changes with a clinician.

References

Xie CQ, Xie FF, Cui JH, Kong LJ, Yao F, Fang M. Physical activity as a protective factor for insomnia: systematic review, meta-analysis, and quality assessment of observational studies. Journal of Global Health. 2026;16:04196. doi:10.7189/jogh.16.04196