Cochrane's updated review finds exercise about as effective as therapy for depression

Cochrane published the seventh version of its review of exercise for depression on January 8 [s1]. The reviews are worth attending to when they update, because the question — does exercise treat depression, and how does it compare with the treatments already in use — is one where the answer has been unstable across two decades of small trials.

This update pools 73 randomised controlled trials with at least 4,985 participants [s1].

Against no treatment: a moderate effect that shrinks under scrutiny

Compared with a control intervention, the pooled standardised mean difference for depressive symptoms at end of treatment was −0.67 (95% CI −0.82 to −0.52), rated low-certainty evidence [s1]. A standardised mean difference of that size is conventionally read as moderate.

The number that deserves equal billing is what happens when the analysis is restricted to trials at low risk of bias. Across 7 such studies with 447 participants, the effect fell to −0.46 [s1].

That drop is the single most informative fact in the review. It is the signature of a literature in which methodological weakness inflates effect sizes — small trials, unblinded outcome assessment, control groups that receive nothing at all rather than an active alternative. Exercise trials are structurally hard to blind, since participants know whether they are exercising. The direction of that bias is predictable.

Against therapy and against medication: little difference, and thin data

Ten trials with 414 participants compared exercise with psychological therapy and found probably little to no difference in effect on depressive symptoms at end of treatment: SMD 0.03 (95% CI −0.16 to 0.23), moderate-certainty evidence [s1].

Five trials with 330 participants compared exercise with antidepressant medication and found there may be little to no difference: SMD −0.11 (95% CI −0.33 to 0.10), low-certainty evidence [s1].

These are the comparisons the headlines are built on, and they are the ones with the least data behind them. Fourteen hundred people would be a modest single trial; here 744 people are split across 15 studies conducted in different populations with different exercise protocols and different comparator therapies.

"No significant difference" in a small comparison is also not the same as "equivalent." The confidence interval around the exercise-versus-therapy estimate spans from a small advantage for therapy to a small advantage for exercise [s1]. What the review supports is that a large difference in either direction is unlikely, not that the two are interchangeable.

Durability is the weakest part

On longer-term effects, the review reports very uncertain evidence: SMD −0.53 (95% CI −1.11 to 0.06), very low certainty [s1]. The interval crosses zero.

Coverage of the review notes that few studies tracked participants after treatment ended [s2]. For a condition characterised by relapse, that is a material gap — arguably the gap that matters most for anyone deciding how to allocate treatment.

What the review says about dose

Reporting on the review describes light to moderate intensity activity as most beneficial, with optimal results between 13 and 36 exercise sessions, and side effects as uncommon [s2].

Those figures come from within-review analyses of trials that varied enormously in what they asked participants to do, and they should be read as descriptive of the evidence base rather than as a prescription. The review's own authors state that future research should prioritise improving trial quality and determining which exercise characteristics work best for different populations [s1].

Other limitations named

Many included studies were small, often involving fewer than 100 participants [s2]. Evidence on which types of exercise suit which populations is insufficient [s2].

What this changes

The reasonable summary is that exercise has a real but probably modest effect on depressive symptoms, that the best available head-to-head data does not show it performing worse than psychological therapy or medication over the treatment period, and that almost nothing is known about what happens after the trial ends.

That is a meaningfully different claim from "exercise is as good as therapy." It is also not nothing: for a low-cost intervention with uncommon side effects [s2], failing to underperform established treatments in the available comparisons is a substantive finding, even at moderate certainty.

What to watch

Whether larger, better-conducted head-to-head trials narrow the confidence intervals; whether follow-up beyond end of treatment is built into new studies; and whether any trial isolates which components of an exercise intervention — the activity, the structure, the supervision, the social contact — carry the effect.

This article describes a published systematic review and is informational only. It is not medical advice and does not recommend any treatment, exercise programme, or change to care.

Sources

• [s1] Clegg AJ, Hill JE, Mullin DS, Harris C, Smith CJ, Lightbody CE, Dwan K, Cooney GM, Mead GE, Watkins CL, Exercise for depression, Cochrane Database of Systematic Reviews, 2026, Issue 1, published 2026-01-08.
• [s2] Scientists find exercise rivals therapy for depression, ScienceDaily, 2026-01-08.