Exercise and Mental Health: What the Evidence Shows

One of the better-evidenced non-drug interventions in mental health, stated at its real strength — neither dismissed nor oversold.

What this is

The strongest recent synthesis is a 2024 BMJ systematic review and network meta-analysis of randomised trials, comparing exercise modalities against each other and against psychotherapy, antidepressants and control conditions for major depressive disorder.

Its findings were specific rather than vague: walking or jogging, yoga and strength training showed the largest effects, with yoga and strength training the best tolerated. It also noted that trials describing participant autonomy tended to show weaker effects, suggesting how exercise is prescribed and supported matters to the outcome.

What the evidence says

  • Exercise reduces depressive symptoms

    Gold· human data

    Supported by a large body of randomised trials, synthesised in the 2024 BMJ network meta-analysis. Effect estimates vary and blinding is inherently difficult, which is why confidence is high but not absolute.

  • Walking, jogging, yoga and strength training performed best

    Gold· human data

    These were the modalities with the largest effects in the network analysis, with yoga and strength training also the most tolerated — useful when energy and motivation are already low.

  • Exercise helps anxiety symptoms

    Gold· human data

    Consistently supported across trials and meta-analyses, though generally with smaller effect sizes than for depressive symptoms.

  • How it's delivered affects how well it works

    Silver· human data

    The BMJ analysis found weaker effects in trials emphasising participant autonomy, implying structure and support are part of the intervention rather than incidental to it.

  • Exercise can replace medication or therapy

    Not supported

    The evidence supports exercise as an option alongside or comparable to other treatments in trials — not as a reason to stop treatment that's working. That decision is clinical.

What this means in practice

  • Pick the modality you'll actually do. Adherence dominates modality choice in practice.
  • Structure helps: a set time, a set place, ideally someone expecting you. That's what the trials delivered.
  • Start smaller than feels impressive. A short walk done daily beats a plan abandoned in week two.
  • Keep any prescribed treatment going, and tell your clinician what you're adding.
  • Strength training and yoga were both effective and well tolerated — worth knowing if running feels out of reach right now.

When to see a doctor

  • Thoughts of harming yourself — contact your GP, NHS 111, or Samaritans on 116 123 now.
  • Symptoms that persist for weeks and affect work, sleep or relationships.
  • Using exercise to punish yourself, or training through injury to avoid rest.

The honest summary

Exercise has genuine, well-analysed evidence for depressive symptoms, with walking, jogging, yoga and strength training leading the field. It belongs alongside other treatment, delivered with enough structure to actually happen.

Work it out

See also

Frequently asked questions

Is exercise as effective as antidepressants for depression?
The 2024 BMJ network meta-analysis compared exercise modalities with antidepressants and psychotherapy and found several exercise types performed well. Direct head-to-head certainty is limited by trial design, and the practical answer is that they're not mutually exclusive — that's a decision to make with a clinician.
Which exercise is best for mental health?
Walking or jogging, yoga and strength training had the largest effects in the network analysis, with yoga and strength training the best tolerated. The one you'll keep doing beats the one that scored marginally higher.
How much exercise do I need for a mood benefit?
Benefits appear across a range of doses, with more vigorous programmes tending to show larger effects in trials. Starting at all is the step that matters most.

Sources

Written and reviewed by Mathew Beale, MSc Biotechnology, University of Reading.

Last reviewed: