"All forms of exercise are as good as, or better than, medication and talking therapies." That line went out in the press materials for a February 2026 paper in the British Journal of Sports Medicine. Notice what it is doing before we get to whether it is true. It is not "may be comparable in some contexts." It is a flat claim of superiority, the kind that writes its own headline, and it is not what the paper underneath it can actually support.
Here is the paper that can support something, quieter and a month earlier. In January 2026, Cochrane, the collaboration whose whole business model is refusing to get excited, published an update to its long-running review on exercise for depression: 73 randomized trials, nearly 5,000 adults with diagnosed depression, led by Andrew Clegg. Against no treatment, exercise produced a moderate reduction in depressive symptoms, though Cochrane rated the certainty of that evidence low. Against psychotherapy, ten trials found similarly sized benefits, at moderate certainty. Against antidepressant medication, the effects looked comparable too, but that comparison rested on evidence Cochrane itself called limited and low certainty, the thinnest leg of the three. As ScienceDaily reported him saying, Clegg's summary was almost defiantly modest: "This suggests that exercise works well for some people, but not for everyone, and finding approaches that individuals are willing and able to maintain is important." Many of the underlying trials were small, often fewer than 100 people, and long-term follow-up was rare, so durability is still an open question. That is a genuinely good story already, a cheap, low-risk intervention holding its own against treatments with real costs and side effects, most solidly against psychotherapy, more tentatively against both no treatment and medication. It did not need embellishing.
The BJSM paper is a different kind of study: an umbrella review, pooling other meta-analyses rather than original trials, a so-called meta-meta-analysis. When the UK's Science Media Centre asked outside experts to weigh in, the pushback was pointed. Dr. Brendon Stubbs of King's College London, who called exercise "a credible, evidence-based option that can sit alongside medication and talking therapies, and in some cases be an effective first step," also noted that 87 percent of the pooled meta-analyses are rated low or critically low quality, and that the claim of exercise beating medication or therapy lacks any head-to-head trial to back it, since the underlying trials never actually pitted exercise against a pill or a therapist in the same study. Professor Jonathan Roiser of University College London went further, flagging inadequate or inappropriate control groups and small samples throughout, and warning that "no one who is currently receiving medication or psychotherapy for anxiety or depression should be stopping their treatment on the basis of this paper." Professor David Curtis, also at UCL, added that the trials only capture people well enough to volunteer for one, so the finding cannot be generalized to depression as such. Professor Anna Whittaker of the University of Stirling struck the most measured note, saying exercise is worth discussing with a doctor as one option among others, not a replacement for the rest.
None of this makes exercise a placebo dressed up as medicine. It means the honest sentence is narrower than the press release: exercise helps, moderately, for a meaningful share of people, with a safety profile medication cannot match, and the data comparing it directly to pills or talk therapy is thinner than "beats" implies. If you are on an antidepressant or in therapy, this is not a cue to stop either one. It is a case for adding a walk, not subtracting a prescription.
