The most successful fitness target in modern history was written by people trying to shift units.

In 1965, a Japanese clock and instrument maker called Yamasa Tokei Keiki released a consumer pedometer named the Manpo-kei, which translates to something like “ten-thousand step meter.” Japan had just hosted the 1964 Tokyo Olympics and was gripped by a national enthusiasm for fitness. The number was round, it was easy to remember, and the kanji for 10,000 happens to look a bit like a person mid-stride. Harvard Medical School epidemiologist I-Min Lee has been unusually blunt about the whole thing, telling Popular Science that no studies had examined 10,000 steps at the time. It sounded good. That was the brief.

Sixty years on, it is the factory default on nearly every wrist in the developed world.

What happened when someone finally checked

Lee ended up running the numbers herself. Her 2019 paper in JAMA Internal Medicine followed 16,741 women with an average age of 72, each wearing a hip accelerometer for a week, then tracked deaths for the next four years or so. Women averaging around 4,400 steps a day had a 41 per cent lower death rate than the least active group, who managed about 2,700, per the National Institutes of Health summary of the work. Mortality kept falling as step counts climbed, then flattened out at about 7,500.

One study, one population, older American women, four years of follow-up. Worth holding loosely.

A far bigger pile of data, much the same answer

A University of Sydney team led by Ding Ding pooled 57 studies across 35 cohorts for a review published last year in The Lancet Public Health, covering mortality, heart disease, cancer, diabetes, dementia, depression and falls. Compared with 2,000 steps a day, 7,000 came with a 47 per cent lower risk of dying from any cause, a 38 per cent lower dementia risk and a 28 per cent drop in falls. The curves bend hardest somewhere between 5,000 and 7,000, then go nearly flat.

Independent physiologists asked to assess the paper by the Science Media Centre mostly agreed with the headline while adding useful friction. Steven Harridge at King’s College London said 10,000 confers little extra benefit for the bulk of the conditions covered, though further risk reduction may still be available for some conditions. Andrew Scott at the University of Portsmouth pointed out that in most cases 10,000 remains better than 7,000, just at a shrinking rate of return. Daniel Bailey at Brunel raised the more awkward limitation, warning that many findings in the review rested on very few studies and that the strength of the evidence reached only moderate for nearly all the outcomes. The paper’s own figures back him up. Its dementia and cancer-incidence estimates each lean on two studies.

Weight is a separate story

This is where the step counter really lets you down.

Researchers at Brigham Young University recruited 120 first-year women and set them a daily target of 10,000, 12,500 or 15,000 steps, six days a week, for 24 weeks, publishing the results in the Journal of Obesity. The 92 who saw it through hit their targets. They also gained weight, roughly 1.5 kilograms on average, with no meaningful difference between the three groups. In coverage by ScienceDaily, lead author Bruce Bailey said tracking steps does lift activity levels but does not translate into holding weight steady.

Over a longer horizon, the picture softens. A 2022 Nature Medicine analysis linked Fitbit records to medical histories for thousands of people in the All of Us Research Program. Obesity risk kept dropping as steps rose, in a straight line instead of a curve, with the sharpest separation above about 8,200 steps a day. Walking shapes body composition on a scale of years. One semester of late-night pizza will outrun it.

What I do with the number now

Most evenings I land somewhere in the high six thousands, ring unclosed, app quietly disappointed in me.

I used to take that personally, like I’d shown up late to my own life. These days I know better. The finish line moved a few thousand steps closer years ago, and nobody bothered updating the app, so it keeps rendering my verdict in red. I’ve stopped reading it as one.