Exercise for Depression and Anxiety: What the Evidence Says

Does exercise actually help depression and anxiety?

Yes, on average, and the effect is not small. The largest overview to date pooled 97 systematic reviews covering 1,039 trials and 128,119 adults. Compared with usual care, physical activity had medium effects on depression (median effect size -0.43), anxiety (-0.42) and psychological distress (-0.60). The benefits held across healthy adults, people with diagnosed mental health disorders and people with chronic disease (Singh et al., 2023).

For people with a clinical diagnosis, a network meta-analysis in The BMJ looked at 218 randomized trials and 14,170 participants with major depression. Compared with active controls such as usual care or a placebo tablet, walking or jogging reduced depression with an effect size of -0.62, yoga -0.55, strength training -0.49, mixed aerobic exercise -0.43, and tai chi or qigong -0.42. The effect appeared similar for people with and without other health conditions, and across different starting levels of depression (Noetel et al., 2024).

The same holds when anxiety is measured directly. A 2026 network meta-analysis of 23 randomized trials and 1,830 adults found exercise improved depressive symptoms (standardized mean difference -0.67) and anxiety symptoms (-0.77) compared with control groups (Zhang et al., 2026).

How strong is the evidence, really?

This is the part most articles skip. Exercise trials are hard to run well: you cannot blind people to whether they are exercising, and most trials rely on self-reported symptom scales. In the BMJ analysis, only one study met the Cochrane criteria for low risk of bias, and confidence in the results was rated low for walking or jogging and very low for the other types (Noetel et al., 2024). In the overview of reviews, 77 of the 97 reviews scored critically low on a standard quality tool (Singh et al., 2023).

The 2026 Cochrane update, which is the most conservative reading of the field, makes the same point with numbers. Across 57 trials and 2,189 participants comparing exercise with no treatment or a control, exercise reduced depressive symptoms (SMD -0.67, low-certainty evidence). When the authors kept only the seven trials with adequate allocation concealment, intention-to-treat analysis and blinded outcome assessment, the effect shrank to -0.46. Long-term follow-up was rare, and the nine trials that had it gave very uncertain evidence (Clegg et al., 2026).

A separate meta-analysis of 41 trials and 2,264 participants found a large effect overall (SMD -0.95), which corresponds to treating about two people for one to benefit. Restricted to the 12 trials at low risk of bias, the effect was moderate (SMD -0.67) and the number needed to treat rose to 2.8 (Heissel et al., 2023). The pattern is the same in every review: the better the trial, the smaller the effect, but it does not disappear.

How does exercise compare with therapy or medication?

In the trials that compared them head to head, exercise performed about as well as either. The Cochrane review found probably little to no difference between exercise and psychological therapy at the end of treatment (ten trials, 414 participants, moderate-certainty evidence), and possibly little to no difference between exercise and antidepressants (five trials, 330 participants, low-certainty evidence). The authors note this rests on a few small trials (Clegg et al., 2026).

One of the more detailed comparisons came from the Netherlands. In 141 outpatients with depression, anxiety disorders or both, 16 weeks of group running therapy at least twice a week produced the same remission rate as escitalopram or sertraline: 43.3% with running and 44.8% with medication. The running group also improved on weight, waist circumference, blood pressure, heart rate and heart rate variability, partly because the medication group got worse on those measures. The limitation matters: most participants chose their treatment rather than being randomized, and far more chose running (Verhoeven et al., 2023).

None of this means exercise should replace treatment that is working. The fair reading is that it belongs alongside therapy and medication as a core option, which is how the BMJ authors framed it.

Which kind of exercise works best, and how hard should it be?

Several kinds work, and intensity seems to matter more than type. In the BMJ analysis, the effects of exercise were proportional to the intensity prescribed, and strength training and yoga were the best tolerated (Noetel et al., 2024). The overview of reviews also found that higher-intensity activity was linked to greater improvements, and that the benefit diminished in longer programs (Singh et al., 2023).

The 2026 anxiety and depression analysis ranked yoga first, followed by moderate continuous training and high-intensity interval training, but the differences between those three were not statistically significant, so the authors called the ranking exploratory (Zhang et al., 2026). In practice, the best type is the one you will keep doing. Supervised and group sessions showed the largest effects in the 41-trial analysis (Heissel et al., 2023), which suggests that structure and other people are part of what works.

Short, hard cardio fits this evidence. Reduced Exertion High-Intensity Interval Training (REHIT) uses two 20-second sprints inside a five-minute ride, which takes time out of the equation. It also targets VO₂max, the most oxygen your body can use, which is one of the strongest measurable markers of longevity. We have not found a trial of REHIT for depression specifically, so treat it as a way to get intensity into your week, not as a proven mental health treatment.

How much exercise do you need?

Less than most people think, especially if you are starting from very little. A dose-response meta-analysis of 15 prospective studies, with 191,130 adults followed for at least three years, found the steepest gains at the lowest activity levels. Adults who did half the recommended weekly volume of physical activity had an 18% lower risk of developing depression than inactive adults. Those who met the full recommendation had a 25% lower risk, with smaller and less certain gains above that (Pearce et al., 2022).

These are observational data, so they show association, not cause. But they agree with the trials on one useful point: the first step from doing nothing to doing something is where most of the benefit appears. The running trial above used at least two sessions a week for 16 weeks, so give a new routine a few months before you judge it. When energy is low, a short session you can repeat beats a long one you dread.

What if exercise has not helped you?

That is a common experience, and the research explains part of it. An average effect across thousands of people says nothing certain about one person. A number needed to treat of about two to three means some people in every trial did not benefit, and in the running study more than half of participants in each group did not reach remission (Verhoeven et al., 2023). Exercise also asks the most of people when motivation is lowest, which is a real barrier, not a personal failing.

If exercise has not helped on its own, it can still help alongside other treatment, and it carries benefits for your heart and metabolism that do not depend on mood. The Cochrane review found adverse events were uncommon, mostly minor musculoskeletal injuries (Clegg et al., 2026).

This article is not medical advice. If you have depression or anxiety, talk to a doctor or mental health professional before changing or stopping any treatment. If you are in crisis, contact local emergency services or a crisis line right away.

The bottom line

Exercise reduces symptoms of depression and anxiety for most people, by roughly as much as therapy or medication in the small number of trials that compared them. The evidence is broad but not high quality, and the effect is smaller in the best-run trials. Walking or jogging, strength training and yoga have the strongest data, harder sessions tend to help more, and the biggest gain comes from going from nothing to something. Expect individual variation, give it a few months, and use it alongside professional care rather than instead of it.