Exercise for Anxiety: What the Research Actually Shows

“Just go for a run” is the most-prescribed and least-examined advice in mental health.

The evidence is better than skeptics claim and weaker than the wellness industry sells. Across meta-analyses, exercise reduces anxiety with a moderate effect — real, repeatable, and on par with the size of effect you’d want from a first-line intervention. But the trials are small, impossible to blind, and the people who benefit most are often the ones who were healthiest to begin with. Here’s what holds up.

How much does exercise actually reduce anxiety?

The cleanest signal comes from people who already have an anxiety diagnosis. A 2017 meta-analysis (Stubbs et al., Psychiatry Research) pooled 6 randomized controlled trials of 262 adults with anxiety or stress-related disorders and found exercise cut anxiety symptoms more than control conditions, with a moderate effect size (SMD = −0.58, p = 0.02).

Zoom out to the whole population and the effect holds but softens. A large 2023 umbrella review (Singh et al., British Journal of Sports Medicine) synthesised 97 reviews covering 1,039 trials and 128,119 participants — the biggest synthesis to date — and found physical activity produced a median effect of −0.42 on anxiety versus usual care. That’s a medium effect, consistent across healthy adults, people with mental-health diagnoses, and people with chronic disease.

So the direction is not in doubt. The magnitude is “helpful adjunct,” not “replaces everything else.”

Does it have to be running — or does lifting count?

It counts. Most of the older evidence is aerobic, but resistance training has its own meta-analysis. Gordon et al. (2017, Sports Medicine) pooled 16 trials and 922 participants and found resistance exercise significantly reduced anxiety (Δ = 0.31, 95% CI 0.17–0.44) — and, notably, the benefit wasn’t moderated by sex, training intensity, or how heavy the program was.

There’s a catch worth sitting with. In that same analysis the effect was larger in healthy participants (Δ = 0.50) than in people with a physical or mental illness (Δ = 0.19). The pattern repeats across the literature: exercise reliably nudges anxiety down, but it does the most work for people who are already relatively well — which is exactly backwards from how it’s usually pitched.

The practical read: pick the modality you’ll actually repeat. Aerobic and resistance both clear the bar, so adherence is the real variable, not the machine.

What’s the minimum effective dose?

This is the part the “do more” crowd gets wrong. In the Singh umbrella review, higher-intensity activity was associated with greater symptom improvement — but effectiveness diminished as interventions ran longer. Short, harder bouts outperformed long, gentle programs. The likely story is dose-response on intensity plus a real-world drop-off in adherence over months, but the signal is consistent: you do not need an endurance habit to get the anxiety benefit.

For a time-boxed high-performer, that’s freeing. A couple of short, genuinely effortful sessions a week is a defensible target — closer to the dose that actually moved the needle in trials than a daily hour you’ll abandon by week three. The lever is intensity and consistency, not volume.

Is exercise as good as medication for anxiety?

Honestly: nobody can tell you, because the head-to-head trials barely exist. The anxiety literature is dominated by exercise-versus-waitlist and exercise-versus-usual-care comparisons, not exercise-versus-SSRI. Anyone quoting you a clean “exercise beats Zoloft” number for anxiety is extrapolating from depression research, where the comparison has been studied more.

Treat exercise as a high-value addition to a plan, not a verified swap for medication or therapy. If you’re on treatment, it stacks; if you’re not, it’s a reasonable first move for mild symptoms — with a low ceiling you should know about going in.

Why don’t the best trials trust these numbers?

Because you cannot blind someone to whether they exercised. That single fact bleeds expectation into every result — people who believe the treadmill will calm them down tend to report that it did.

A pointed 2024 critical review (Stonerock, Gupta & Blumenthal, Progress in Cardiovascular Diseases) screened the recent RCTs and found that of 13 trials that actually enrolled people with elevated anxiety at baseline, only 2 showed anxiety was unequivocally reduced by exercise. Most studies, they noted, were undermined by concurrent therapies, missing intention-to-treat analysis, and the unavoidable absence of blinding. Add publication bias — null results on exercise rarely get written up — and the pooled effect sizes are probably the optimistic edge of the truth.

None of that means exercise doesn’t work. It means the confident numbers deserve an asterisk, and the mechanism may be as much about agency, routine, and downregulated arousal as anything metabolic. If your anxiety is body-led — a nervous system stuck in high gear — movement is one input among several; why your body keeps the score and what the breathwork evidence actually supports cover the others.

The takeaway

Exercise lowers anxiety with a moderate, well-replicated effect — strongest in people who are already fairly well, real but oversold in people who aren’t. Favor intensity over duration, pick the modality you’ll repeat, and read the glossiest claims with the blinding problem in mind. It’s one of the best low-risk levers you have; it just isn’t the only one. For how movement fits the larger system of winding a nervous system down, see our anxiety regulation and sleep restoration work.