In 1965 a Japanese company called Yamasa Tokei Keiki began selling a pedometer. They named it the Manpo-kei, which means "ten-thousand-step meter". The number was chosen because it was round, because it sold, and because the Japanese character for ten thousand looks something like a person walking.

No study had examined ten thousand steps at the time. There was nothing to examine. The figure that now sits as the default goal in most fitness trackers began as the name of a product, in the years after the 1964 Tokyo Olympics, when Japan was worried its citizens had stopped moving.

Sixty years later the research has caught up, and the interesting part is not that the number was invented. It is how close the invention landed.

What the evidence actually supports

The largest synthesis to date was published in The Lancet Public Health in 2025, led by Melody Ding at the University of Sydney. It pooled 57 studies conducted between 2014 and 2025 across more than ten countries, and it compared people walking 7,000 steps a day with people walking 2,000.

outcome                 risk at 7,000 steps vs 2,000
all-cause mortality              47% lower
dementia                         38% lower
type 2 diabetes                  27% lower

The same pattern held for cardiovascular disease, cancer, depression and falls. Seven thousand, not ten, is where the review lands as a target that is both supported and achievable.

The shape matters more than the number

The dose-response curve is steepest between 2,000 and somewhere around 5,000 to 7,000 steps. After that it flattens. It does not turn back on itself, and nothing in the data suggests that walking more is harmful — the extra steps simply buy less than the earlier ones did.

Dementia is a clear illustration. Going from 2,000 steps to 7,000 is associated with 38% lower risk; the next 3,000 steps, from 7,000 to 10,000, adds about another 7% on top of that in the reviewers' own wording. The first big push does most of the work.

The practical conclusion, though, is not the one the debunking usually reaches. If you are currently sedentary, the steps that matter most are the ones you are not taking yet. Getting from 2,000 to 5,000 is worth more than getting from 7,000 to 10,000, and it is a great deal easier.

Two earlier studies had already found the same shape

The 2025 review confirmed a pattern seen in earlier work.

In 2019, I-Min Lee and colleagues published a study in JAMA Internal Medicine following about 17,000 women with a mean age of 72 for over four years. Women averaging 4,400 steps a day had 41% lower mortality than those averaging 2,700. The benefit kept improving with more steps and then levelled off at roughly 7,500.

That study also tested something people assume matters: pace. Once total daily steps were accounted for, how fast those steps were taken showed no association with mortality. The volume carried the effect, not the speed.

In 2022, Amanda Paluch and colleagues pooled 15 cohorts covering 47,471 adults and 3,013 deaths, and found the plateau sat at 6,000 to 8,000 steps for people aged 60 and over, and 8,000 to 10,000 for those under 60. Older adults reach the flat part of the curve sooner, which is worth knowing if you are setting a goal for a parent rather than for yourself.

What the debunking gets wrong

"Ten thousand steps is a myth" has become its own piece of folklore, and it is usually deployed to mean that step counts do not matter. The evidence does not support that either.

The number was arbitrary in origin and sits about 43% above where the benefit curve flattens. Both of those things are true, and neither makes walking optional. A 47% difference in all-cause mortality between 2,000 and 7,000 steps is not a rounding error, and it is one of the few health findings that replicates across cohorts, countries and decades.

The correct reading is narrower and duller than either slogan: the target was made up, the real figure is lower, and the direction was right.

What a step count cannot see

Steps are a crude instrument, and their main flaw is what they miss.

They do not capture intensity. They represent cycling, swimming and rowing badly or not at all, so an hour in a pool registers as a sedentary hour. And a device on your wrist counts differently from one on your hip, which is part of why step research is harder to pool than it looks.

That is not an argument against counting steps, but against treating the count as a complete measure of your day.

Where this comes from, and what it does not tell you

The 2025 figures are from Ding and colleagues in The Lancet Public Health; the 2019 study is Lee and colleagues in JAMA Internal Medicine; the 2022 pooled analysis is Paluch and colleagues, also in The Lancet Public Health. The origin of the Manpo-kei is well documented and the observation that no contemporaneous study supported the number is I-Min Lee's own. Read on 30 August 2026.

This is not medical advice. It is a description of what a body of research reports about a population. If you have a heart or joint condition, are pregnant, or are recovering from anything, the number that applies to you is the one your clinician gives you, and it may be higher or lower than any figure here.

The studies are observational. They record what people did and what happened next; they did not assign anyone to walk. People who walk more differ from people who walk less in many ways, and although the analyses adjust for a great deal, reverse causation is a live concern in this literature — early illness reduces walking before it shows up as a diagnosis.

The evidence is rated moderate, not strong. Several of the outcomes rest on a small number of studies, and the confidence intervals around specific figures are wider than a single headline percentage suggests.

The populations were generally healthy. The reviewers say plainly that the findings do not transfer to people living with a chronic condition, which is a large share of the people most interested in the question.

Seven thousand is a marker, not a magic threshold where benefits suddenly switch on. Risk falls with roughly each additional thousand steps, and a goal anywhere on that curve is a reasonable goal. The value of the figure is that it is achievable for people the 10,000 target quietly discourages.