In the Harvard Aging Brain Study, a few hundred volunteers strapped small step counters to their waistbands and then carried on with their lives. They were between 50 and 90, and none had memory problems when they joined. Over the following years, researchers tracked their thinking skills every year, and for a subset they also scanned the brain for two proteins linked to Alzheimer’s disease, amyloid and tau. When the team at Mass General Brigham published the results in Nature Medicine in November 2025, the finding that travelled furthest was a number: roughly 3,000 steps a day. The study has been back in the news over the past couple of weeks, and it deserves a careful reading rather than a headline one.
What the researchers actually did
The analysis covered 296 adults. At the start, PET scans showed that some of them already had elevated amyloid in the brain, the sticky plaque that builds up years before symptoms appear. Others had low amyloid. Everyone’s daily steps were counted with a pedometer, and the team followed them for an average of about nine years, with some people watched for as long as 14.
The interesting results came from the group with elevated amyloid, since these are the people thought to be at higher risk of later decline. Within that group, people who walked between about 3,000 and 5,000 steps a day showed a slower drop in thinking and memory scores, delaying noticeable decline by roughly three years on average compared with the least active people. Those walking about 5,000 to 7,500 steps a day did better still, with an estimated delay of around seven years. Past roughly 7,500 steps the extra benefit seemed to level off.
Tau told the other half of the story. In people with elevated amyloid, higher step counts went along with slower build-up of tau, the protein tangles that track more closely with memory loss. The authors’ modelling suggested that much of the cognitive benefit was linked to that slower tau accumulation. Amyloid levels themselves did not appear to change with activity.
Why 3,000 is not a magic number
It is tempting to read “3,000 steps” as a prescription. It isn’t one. The figure marks the lower edge of the band where researchers saw a difference, and the comparison group was people who moved the least, those under about 3,000 steps. For anyone who currently manages 1,500 on a sitting-heavy day, that is a more realistic first target than the usual 10,000, which has never had a firm scientific basis anyway. Three thousand steps is around 25 to 30 minutes of easy walking, or a mile and a half for many people.
Notice, too, that the benefit curve was not a straight line. Going from very little to a modest amount appeared to matter most. That fits what other walking research has suggested for years: the first few thousand steps seem to carry the largest share of the gain, and each additional thousand adds a little less.
If you want ideas for packing more movement into a day without a gym, our guide to losing weight without a gym has several that work just as well for steps. Anyone who wants to make a short walk more demanding can look at Japanese interval walking, although the Alzheimer’s study did not test intensity at all.
What the study does not show
This is where the headlines tend to run ahead of the science, so it is worth being plain about it.
The study was observational. Nobody was assigned to walk more or less. People who walk more tend to differ from people who walk less in many ways: they may be healthier at the start, eat differently, sleep better, smoke less, or have fewer heart problems. Statistical adjustment can account for some of these differences, but never all of them. Commentators have also pointed out that the opposite direction is possible. Someone in the very early stages of brain change may begin to walk less, so low steps could be a sign of what is coming rather than a cause of it.
The participants were also a particular group: 296 people from one research cohort in the United States, described by outside experts as mostly white and well educated. Results may not carry over to every population, and the study was too small to say much about individuals. A pedometer counts steps but not pace or effort, so the findings say nothing about whether brisk walking beats strolling. And in people whose amyloid was low, the team did not find a meaningful link between steps and decline, partly because those people showed very little decline to begin with.
Finally, this is about slowing a decline in people already showing early biological signs. It does not show that walking prevents Alzheimer’s disease, treats it, or reverses it. Dementia has many contributors, including age, genetics, hearing, blood pressure, and education, and a step count touches only some of them.
Putting it into an ordinary week
A sensible reading is modest. Moving more is linked with better brain outcomes in this and several other large studies, it also supports the heart and mood, and it carries little risk for most people. Our piece on strengthening your heart naturally covers the cardiovascular side, which overlaps considerably with brain health.
If your current average is low, find it first. Most phones record steps without any setup. Look at the past two weeks, take the average, and aim to add about 1,000 steps a day to it. A ten-minute stroll after lunch typically supplies that much. When that feels easy, add another walk, perhaps a phone call taken on your feet or one stop earlier on the bus. People who already reach 5,000 to 7,500 steps are, on this evidence, in the zone where the study found its largest effect, and there is no sign that grinding out 15,000 would be better for the brain.
Where you walk matters for comfort and safety too. On days when the air is smoky or very hot, a mall or an indoor track will do. In autumn and winter, wet leaves and early dark raise the risk of a fall, so a lit, even route and sturdy shoes are worth the trouble. People with heart conditions, joint problems or balance issues should check with their doctor before changing how much they walk.
One last point is worth keeping in mind. The research team has said it plans to look next at activity patterns and intensity, which a pedometer cannot capture. Until trials and further studies arrive, the honest summary is this: more walking is associated with slower decline in people at risk, the gains looked biggest between 3,000 and 7,500 steps a day, and nobody has yet proven that the steps themselves are responsible.
This article is for general information and is not medical advice. If you are worried about memory changes in yourself or someone close to you, talk to a doctor.