Walking and dementia: how many steps to delay it, and by how many years

If there is one intervention I wish I could prescribe to every patient over fifty, it would be regular walking. Not because it’s sophisticated. Because the evidence on what it does to the brain keeps getting stronger.

Three studies published in 2025 put fresh numbers on the question. The findings converge on the same message. The floor for meaningful benefit is lower than most guidelines suggest, and the effect sizes at genuinely modest levels of activity are larger than most people assume.

The Nature Medicine amyloid cohort

The most interesting of the three is a prospective cohort study published in Nature Medicine, which followed 296 adults over several years.¹

What made this cohort unusual was the inclusion criterion. These participants had elevated brain amyloid, the early biochemical signature of Alzheimer’s risk, measured on imaging. So the question wasn’t whether exercise reduces general dementia risk in healthy people. It was whether exercise slows cognitive decline in people already showing preclinical pathology.

The finding was a clear dose-response between daily step counts and preservation of cognitive function. Higher daily activity was associated with substantial delays in measurable cognitive decline, even within this amyloid-positive group.

This is important. Exercise doesn’t reverse Alzheimer’s pathology. But in people with the underlying biology already in motion, physical activity appears to shift the timeline meaningfully.

The JAMDA “minimum effective dose” finding

The second study, published in the Journal of the American Medical Directors Association in 2025, asked a different question. What is the smallest amount of moderate-to-vigorous physical activity associated with a meaningful reduction in dementia risk?²

The answer surprised me when I first read it.

1 to 35 minutes per week of moderate-to-vigorous activity (so as little as five minutes a day) was associated with a 41% lower risk of all-cause dementia compared with doing none. And critically, this benefit held even in older adults classified as frail.

Let me be clear about what this does and doesn’t mean. It does not mean five minutes a day is the target. More activity adds more benefit, up to a point. What it means is the floor for any measurable benefit is extremely low. You do not have to hit 150 minutes a week before exercise begins protecting your brain. The very first minutes count.

This finding matters most for people who feel daunted by conventional exercise guidelines and therefore do nothing. The honest message to those patients is: do anything. Do a five-minute walk. Park further away. Climb one flight of stairs. The research says the return on that small investment is real.

The meta-analysis on Alzheimer’s specifically

The third study, a 2025 dose-response meta-analysis in the Journal of Neurology, pooled prospective cohort data to quantify the relationship between physical activity and Alzheimer’s disease specifically (not all-cause dementia).³

High physical activity was associated with roughly a 26% reduction in Alzheimer’s risk compared with low activity.

The dose-response curve was curvilinear. Steepest between low and moderate activity, flattening as activity levels rose. This is the same shape we’ve seen for cardiovascular mortality: the biggest gains come from moving out of the sedentary category, not from pushing from moderate to extreme activity.

Why does exercise protect the brain?

Multiple mechanisms, acting in parallel:

• Cerebral blood flow improves with regular activity, which supports neuronal function and reduces the vascular contribution to cognitive decline.
• Chronic inflammation drops, which is implicated in amyloid and tau pathology.
• BDNF (brain-derived neurotrophic factor) rises with exercise. BDNF supports the growth and survival of neurons, particularly in the hippocampus, the region most affected early in Alzheimer’s.
• Insulin sensitivity improves. Insulin resistance in the brain is increasingly recognised as a contributor to Alzheimer’s risk.
• Sleep quality improves, and sleep is when the glymphatic system clears metabolic waste (including amyloid) from neural tissue.

None of these mechanisms is dominant. They stack. Regular physical activity is probably the single most broad-spectrum intervention we have for brain health, because it influences so many of the pathways that drive dementia.

What kind of activity counts?

The research base is largest for aerobic activity: walking, cycling, swimming. The JAMDA study specifically measured moderate-to-vigorous activity, which at population level is mostly brisk walking.

Dance and racket sports add cognitive and social demand on top of physical activity, and they consistently show strong signals in dementia cohorts too. Resistance training protects muscle and bone, which matters for fall prevention and therefore for maintaining independence, and it may have additional cognitive benefit through its effects on insulin sensitivity and growth factor signalling.

At a practical level, what matters most is consistency. Any regular movement beats infrequent intense effort for cognitive outcomes. Walking is the most accessible and sustainable option, which is why I spend so much time in clinic on it.

What I tell patients

Three things.

The floor is low. If you have done essentially no exercise for years and feel overwhelmed by the standard advice, the research says five minutes a day of moderate activity is already moving the needle. Start there. The first step is bigger than the second hour.

The biggest gains are at the bottom of the dose-response curve. Going from sedentary to walking 30 minutes a day produces a larger reduction in dementia risk than going from 30 minutes to 60 minutes.

Consistency beats intensity. The brain responds to regular stimulus, not heroic effort. Most days of the week, modest amounts, maintained for years.

The bottom line

The 2025 data has made the case for walking as dementia prevention about as evidence-based as it gets. The floor for benefit is extremely low. The dose-response is real. The effect holds even in people with preclinical pathology and in older adults with frailty.

If you’re waiting for a pharmaceutical breakthrough in dementia, you may be waiting a long time. The intervention that already has the strongest evidence is probably in your shoes, and starts the moment you put them on.