Exercise for Anxiety: What the Trial Evidence Actually Shows

Exercise is probably the most universally recommended lifestyle intervention for anxiety. And the recommendation is reasonable — exercise has broad health benefits and is unlikely to cause harm. But when you look at the trial evidence specifically for anxiety, it’s thinner and more complicated than the confident recommendations suggest.

What meta-analyses find

A meta-analysis of RCTs in people with diagnosed anxiety disorders found a significant anxiolytic effect of exercise, with a moderate effect size [1]. But the review included only 12 trials with 672 participants total — a small evidence base for such a universal recommendation.

A broader systematic review of 37 trials found a similar moderate effect, but rated the overall study quality as low [6]. A meta-analysis specifically of resistance training found significant anxiety reduction across 16 trials, though again with small samples and methodological concerns [2].

The consistent finding is a moderate benefit, but with consistent caveats about evidence quality. For what effect sizes mean in practice, see understanding effect sizes.

Why the evidence is weaker than expected

Given that exercise is widely available, cheap and has been studied for decades, you might expect robust, conclusive evidence. Several factors explain why it’s not:

Blinding is impossible. People know whether they are exercising. This means expectation effects are uncontrolled, just as in mindfulness trials. Someone assigned to an exercise program expects to feel better; someone on a waiting list does not. This asymmetry inflates the apparent benefit.

Control conditions are weak. The largest effects appear in comparisons with waiting lists or treatment as usual. When exercise is compared with active controls (stretching, social interaction, health education), the difference shrinks [6]. This suggests that at least some of the benefit comes from structure, attention and expectation rather than exercise itself.

Trials are small. Many exercise trials for anxiety have 20–40 participants per arm. This makes them vulnerable to the same statistical problems that affect other small anxiety trials.

Adherence is a problem. Exercise requires ongoing effort. Dropout rates in exercise trials are often high, and those who complete the program may be systematically different from those who drop out — more motivated, less severely affected, or both. This biases the results in favor of exercise, the same dropout problem that affects many treatment trials.

Outcome timing. Most trials measure outcomes at the end of the exercise program. Whether the benefit persists after someone stops exercising is poorly studied. If exercise needs to be maintained for ongoing benefit, that has different implications from a time-limited treatment like CBT.

The mechanism question

How exercise might reduce anxiety is not established. Several mechanisms are proposed:

• Biological: Exercise increases serotonin, endorphins and brain-derived neurotrophic factor. It reduces cortisol and inflammation. These are plausible pathways but are mostly demonstrated in animal models or measured indirectly in humans.
• Distraction: Exercise provides a break from worry and rumination. But so do many activities.
• Self-efficacy: Completing challenging exercise may build a sense of capability. This is theoretically plausible but hard to isolate.
• Behavioral activation: For people who are sedentary and avoidant, exercise gets them out and active. The benefit may be partly about breaking avoidance patterns.
• Social contact: Group exercise involves being around people, which itself may help.

Any or all of these could contribute, and their relative importance may vary between individuals [4]. The mechanism question matters because it affects what kind of exercise might be recommended and whether similar benefits could come from other activities.

What’s missing from the evidence

Dose-response data. Whether 20 minutes three times a week is different from 45 minutes five times a week is poorly studied. Most trials pick one dose and test it; systematic dose comparisons are rare.

Head-to-head comparisons with standard treatment. Trials comparing exercise directly with CBT or medication are few and small. Claims that exercise is “as effective as antidepressants” rest on very limited direct evidence.

Population specificity. Whether exercise works differently for GAD versus panic disorder versus social anxiety has barely been studied.

Questions to ask about exercise-anxiety claims

1. Was the comparison group a waiting list or an active control?
2. How many participants completed the trial?
3. Were participants diagnosed with an anxiety disorder or from the general population?
4. Was follow-up measured after the exercise program ended?
5. How was exercise adherence measured and reported?

What remains uncertain

Exercise is a reasonable thing to recommend for anxiety, with the honest caveat that the specific evidence is weaker than the strength of the recommendation implies. Whether specific types, intensities or durations are optimal is unknown. Whether exercise can serve as a standalone treatment for anxiety disorders or is better positioned as an adjunct to established treatments is an open question. The most honest summary is that exercise probably helps, but by how much and through what mechanism we cannot confidently say.

Thinking about exercise for anxiety. If you want to try exercise for anxiety, the evidence supports regular moderate activity of a type you can sustain. It’s not a guaranteed fix, and it shouldn’t replace treatment if your anxiety is severe. If anxiety is affecting your daily life, talk to a GP or mental health professional. If you are in crisis or thinking about harming yourself, contact emergency services or a crisis line now: in the US, call or text 988; in the UK and Ireland, Samaritans on 116 123.