Does training actually improve sleep quality?

Yes — across a wide range of studies, structured exercise is consistently linked to better sleep quality, measured by scales such as the Pittsburgh Sleep Quality Index (PSQI). The effect shows up across aerobic exercise, resistance training, yoga, Pilates and combined programmes, though the size and the specific sleep problem it helps with varies by exercise type.

What the research actually looked at

The largest piece of evidence here is a network meta-analysis of 86 randomised controlled trials involving 7,276 people, comparing aerobic exercise, resistance training, combined aerobic-resistance work, yoga, Pilates and traditional Chinese sports such as tai chi (systematic review, 2025). All six categories produced a statistically significant improvement in sleep quality compared with no intervention. Pilates came out ranked highest in this analysis, followed by aerobic exercise, then combined training and resistance training, with yoga and traditional Chinese sports somewhat lower — though all were better than doing nothing. A separate network meta-analysis focused specifically on middle-aged and older adults, using PSQI data from 3,460 participants, found aerobic exercise most effective for total PSQI score and for reducing the time it takes to fall asleep, while yoga performed best for sleep disorders, sleep efficiency, sleep duration and daytime function (systematic review, 2024). So the "best" exercise type depends partly on which sleep problem is being targeted — falling asleep versus staying asleep versus feeling rested during the day.

A third network meta-analysis, looking at non-elderly adults specifically, identified resistance training as the most effective single intervention for sleep quality among the non-drug options it compared, alongside general physical activity and nutritional approaches (systematic review, 2024). Taken together, these three reviews do not agree on a single winner, which suggests the effect of exercise on sleep is broad rather than tied to one narrow protocol.

Does it work on its own, or only alongside other things?

A randomised trial of 112 sedentary women aged 18 to 30 with poor sleep tested high-intensity circuit training alone, a structured sleep health programme alone, the two combined, and a control group over eight weeks (controlled trial, 2026). All three active groups improved on sleep efficiency, sleep latency, time awake after falling asleep, and total PSQI score compared with control. But the combined exercise-plus-sleep-programme group did better than either approach alone on objective measures such as sleep efficiency and reduced night-time waking, measured by wrist actigraphy rather than self-report alone. This points to exercise and behavioural sleep strategies working through somewhat different pathways, with combining them producing a larger effect than either on its own. Readers wanting more on the behavioural side specifically may find what sleep hygiene habits actually work useful.

A smaller trial in adults with type 2 diabetes, aged 55 to 75, compared aerobic exercise, resistance exercise, combined exercise and a control group over 12 weeks (controlled trial, 2025). All three exercise arms improved PSQI score, but only the combined aerobic-plus-resistance group improved objectively measured total sleep time, sleep efficiency and wake time after sleep onset. This is a population with a specific metabolic condition, so it does not necessarily generalise, but it echoes the pattern that combined training may reach outcomes that a single exercise type does not.

How much and what kind — is there a right amount?

The large network meta-analysis of 7,276 participants also modelled a dose-response relationship and described it as U-shaped rather than straightforwardly "more is better" (systematic review, 2025). It reported an overall optimal range around 920 MET-minutes per week, with the ideal dose varying by exercise type — lower for Pilates, higher for aerobic exercise. This archive does not give intake or exercise schedules, but the underlying finding — that benefit does not simply increase indefinitely with more exercise — is worth knowing before assuming that harder or longer training automatically means better sleep. For a broader look at how much movement is linked to sleep benefit, see how much exercise do I need to sleep better and which type of exercise helps sleep the most in midlife.

How exercise compares with other approaches

Cognitive behavioural therapy for insomnia (CBT-I) is a separate, well-studied non-drug treatment, distinct from exercise. A meta-analysis of insomnia comorbid with psychiatric and medical conditions found CBT-I produced medium to large improvements in sleep efficiency, time to fall asleep, and subjective sleep quality (systematic review, 2015), and a systematic review commissioned by the American Academy of Sleep Medicine reached broadly supportive conclusions about CBT-I and related behavioural techniques for chronic insomnia (systematic review, 2021). These are not exercise trials, but they matter for context: exercise is not the only lever available, and in the combined trial above, a sleep-focused behavioural programme performed comparably to circuit training on many measures. For a wider view of this territory, the Sleep, energy and the brain archive covers both angles.

One further data point comes from outside general population research: a trial in gastric cancer patients found a combined programme of CBT, resilience training and sleep hygiene education improved PSQI scores alongside psychological and recovery outcomes after surgery (controlled trial, 2026), and a separate review looked at what helps sleep specifically after cardiac surgery (see what helps you sleep after heart surgery). These are clinical, post-surgical contexts rather than everyday midlife sleep, so they should be read as showing that structured behavioural-plus-sleep programmes have a role in recovery settings, not as direct evidence about exercise for someone otherwise healthy.

What the archive does not answer

None of the trials given here focus specifically on women in midlife or perimenopause — the largest exercise trial used women aged 18 to 30, and the network meta-analyses pool mixed-age and mixed-sex populations without breaking results out by menopausal status. The archive also has nothing here connecting exercise-driven sleep improvement to specific supplements such as melatonin, magnesium or ashwagandha — those are covered separately in their own evidence pages. And while sleep problems in demanding occupations, including women in the military, are documented elsewhere in this archive, the exercise-specific trials here were not conducted in that population. Finally, most of the studies rely on self-reported sleep quality (PSQI) rather than objective measurement, so where a trial did use actigraphy — as the combined circuit-training study did — that distinction is noted above because it strengthens what is being described, not because the archive ranks it as better evidence.