Exercise is sold as brain insurance. The trial evidence is weaker than the headlines.

Physical activity is the closest thing there is to a consensus lifestyle recommendation for the ageing brain, and there are excellent reasons to be active that have nothing to do with dementia. But the specific claim that exercise protects cognition is on softer ground than its ubiquity suggests. The observational evidence is strong; the randomised evidence — the kind that can separate cause from correlation — is mixed, and at its most rigorous it has come up empty.

Why the observational case is not enough

People who are fitter and more active in midlife and later tend to have better cognition and lower dementia rates, and physical inactivity is one of the 14 modifiable risk factors in the Lancet Commission's dementia model. The problem is the same one that dogs most lifestyle-and-brain research: the people who exercise differ systematically from those who do not, in education, income, other health behaviours and baseline health, and those differences are hard to remove statistically. Worse, the earliest phase of cognitive decline itself tends to reduce activity years before diagnosis — so some of the association runs backwards, with declining brains producing declining exercise rather than the reverse.

Randomisation cuts through that. Assign people to exercise or not, follow their cognition, and confounding and reverse causation largely disappear. This is where the picture gets more sober.

The rigorous null

A Cochrane review examined randomised controlled trials of aerobic exercise for cognition in older people without known cognitive impairment [s1]. Its conclusion was blunt: it found no evidence in the available trial data that aerobic activities — including those that successfully improved cardiorespiratory fitness — had any cognitive benefit in cognitively healthy older adults [s1]. In other words, the trials that got people genuinely fitter did not, on average, make them measurably sharper. The review called for larger studies to test possible moderators before drawing firmer conclusions [s1].

That is a narrower finding than "exercise does nothing for the brain." It concerns aerobic exercise specifically, cognitively healthy participants specifically, and the trials available to that review. But it is a direct contradiction of the marketing, and it comes from the most methodologically conservative corner of the evidence base.

The more optimistic meta-analysis, and its limits

A broader systematic review in the British Journal of Sports Medicine took in more than just aerobic training. Pooling 333 effect sizes from 36 randomised trials in adults over 50, it found that exercise improved cognitive function overall, with a standardised mean difference of 0.29 (95% confidence interval 0.17 to 0.41) [s2]. Aerobic exercise, resistance training, multicomponent training and tai chi each showed significant point estimates, and sessions of 45 to 60 minutes at least moderately intense were associated with benefit [s2].

An effect size of 0.29 is small-to-moderate — real, but not the transformation implied by "exercise your way out of dementia." And the two reviews are less contradictory than they look: the broader one includes resistance and mind-body training and a wider outcome set, which is part of why it finds a signal the aerobic-only review did not. It is fair to read them together as: across many kinds of exercise, there is probably a small average cognitive benefit, and for aerobic exercise judged by the strictest standard, that benefit was not demonstrable [s1] [s2].

What this leaves a reader with

The honest position is unglamorous. Exercise has robust, well-established benefits for cardiovascular health, strength, mood, sleep and independence — and several of those, particularly blood-pressure and vascular health, are themselves linked to lower dementia risk. Movement is one of the best-supported things a person can do for overall health in later life. What the trial evidence does not support is the tidy promise that a given amount of exercise buys a specific reduction in your personal cognitive decline. The measurable cognitive effect in randomised trials is small on average and, for aerobic exercise in healthy older adults tested rigorously, was not there at all [s1] [s2].

None of that is a reason to sit still. It is a reason to be active for the many things exercise clearly does, and to treat the "brain insurance" framing as an overstatement of a modest and uneven effect rather than the settled fact it is usually presented as.

This article is informational and is not medical advice.