Half of exercise's brain benefit may travel through better sleep

Exercise reliably improves cognition in older adults, but nobody fully understands how the benefit is delivered. Sleep is a plausible courier: sleep disturbances affect nearly half of older adults and are themselves linked to cognitive decline. If a meaningful share of exercise's brain benefit actually runs through improved sleep, that changes how interventions should be designed and evaluated, because sleep becomes a target in its own right rather than a side effect. This analysis set out to test, formally, whether sleep quality mediates the relationship between exercise and cognition in people aged 60 and over.

The researchers searched five databases, PubMed, Web of Science, Cochrane CENTRAL, Embase and PsycINFO, from inception to June 2026. They included randomised controlled trials of structured exercise lasting at least 4 weeks in adults aged 60 or older, with the unusual requirement that each trial reported both cognitive and sleep outcomes. Eight studies covering 841 participants qualified. Random-effects meta-analysis estimated the overall effects, and meta-analytic structural equation modelling tested whether the sleep improvement statistically carried part of the cognitive gain. Subgroups examined baseline cognitive status and intervention duration.

Exercise significantly improved cognitive function, with a large pooled effect of 1.32, and significantly improved sleep quality, with an effect of 1.13. The mediation model found sleep quality was a significant partial mediator, with an indirect effect of 0.300, accounting for 50.8% of the total cognitive benefit. The split by cognitive status was striking: sleep carried 67.6% of the benefit in people with mild cognitive impairment, against 31.9% in cognitively healthy older adults. The mediation was partial, meaning exercise retained a direct effect on cognition beyond the sleep pathway.

Most reviews in this area establish that exercise helps cognition; far fewer ask how. This analysis moves the question from whether to mechanism, and it nominates sleep as a candidate carrying roughly half the effect. The estimates rest on only eight trials and on statistical modelling rather than trials designed to test mediation, and the authors are explicit that larger, purpose-built studies are needed to confirm them. It also leaves open why the sleep pathway appears so much stronger in people with mild cognitive impairment than in healthy peers.

For an older reader, or anyone thinking ahead, this is a reason to see exercise and sleep as one intertwined system rather than separate boxes to tick. If these exploratory estimates hold up, part of what exercise does for the ageing brain may be delivered overnight, through sleeping better. It is informational rather than a prescription: the trials tested structured exercise programmes, and the mediation figure is a statistical attribution, not a demonstration that fixing sleep alone produces the same cognitive gain.