Neither the World Health Organization nor the US physical activity guidelines have ever recommended a daily step count. Every official target is written in minutes. Into that gap walked a 1965 product name — and it has held the field for sixty years.
Check your phone. It counts steps. Check your watch. Steps. Check any official physical activity guidance in the world, and you will find no step target at all.
The WHO’s 2020 guidelines recommend 150 to 300 minutes a week of moderate activity, plus strength work twice a week. The US Physical Activity Guidelines say the same. Neither mentions a step goal. In the American guidelines the number 10,000 appears once in the main text — in a note that walking that much does not appear to worsen osteoarthritis. A safety observation, not a target.
So every device measures a unit no health authority has ever set a goal for, and every health authority sets goals in a unit no device displays. That translation gap is the real story here, and something did fill it.
Where the number came from — carefully
The standard account is that a Japanese company, Yamasa, launched a pedometer in 1965 called the manpo-kei, which translates as “10,000 steps meter.”
That account is probably right, but it deserves more honesty than it usually gets. It entered the scholarly record through a 2004 review by Catrine Tudor-Locke and David Bassett, and its single documentary citation is a 1993 article by Dr Yoshiro Hatano in a small physical-education journal, plus a conference talk Hatano gave in 2001. Later papers repeat the story citing Tudor-Locke rather than Hatano.
Nobody has published the 1965 advertisement. The origin story rests on a citation chain with one root, and that root is difficult to obtain. It is fair to say the figure is “traceable to” a pedometer brand name. It is not fair to say this is thoroughly documented.
It is also a mistake to conclude that the number is therefore worthless, which is where most debunking goes wrong.
What the dose-response evidence shows
The most recent and most careful synthesis is a 2025 meta-analysis in The Lancet Public Health, covering 57 studies from 35 cohorts. Comparing 7,000 steps a day against 2,000, it found:
- All-cause mortality: hazard ratio 0.53 (95% CI 0.46–0.60)
- Cardiovascular mortality: 0.53 (0.37–0.77)
- Dementia: 0.62 (0.53–0.73)
- Cancer mortality: 0.63 (0.55–0.72)
- Depressive symptoms: 0.78 (0.73–0.83)
- Type 2 diabetes: 0.86 (0.74–0.99)
The authors describe 7,000 steps as “a more realistic and achievable target,” while noting 10,000 “can still be a viable target for those who are more active.” Their curves show an inflection around 5,000 to 7,000 for mortality, cardiovascular disease, dementia and falls — but importantly, no plateau at all for cardiovascular mortality, cancer mortality, diabetes or depression. For those, more kept helping.
A 2023 meta-analysis of 226,889 people found every additional 1,000 steps a day associated with a 15 per cent lower mortality risk (HR 0.85, 0.81–0.91). The age split matters: maximum benefit came at 6,000–10,000 steps for over-60s and 7,000–13,000 for under-60s.
And the largest single study — 78,500 UK Biobank participants — found benefits continuing up to about 10,000. A companion paper put the optimal dose for avoiding dementia at 9,826 steps a day.
So 10,000 is not a myth. It is a defensible number for a healthy adult under 60. It is simply not the number, and it is close to the worst possible number to give the people with most to gain.
Why the target is aimed at the wrong people
Here is the awkward fact. The average person on Earth walks 4,961 steps a day — measured across 68 million days of smartphone data from 717,527 people in 111 countries. In the United States it is 4,774. In the UK, 5,444.
The default target is therefore roughly double what a typical person currently does. And the evidence says the steepest returns are at the bottom of the range, not the top: mortality risk becomes measurably lower from about 2,517 steps a day, and half the maximum dementia benefit is reached by 3,826.
Someone walking 3,000 steps who is told the goal is 10,000 is being handed a target more than three times their current life, at the exact point on the curve where a few hundred more steps would do them the most good.
The evidence on round-number targets
This is not speculation. Two randomised trials have tested how the goal is set.
The first randomised 96 adults to either a static 10,000-step goal or an adaptive one that adjusted to recent performance. The static goal produced a bigger initial jump — and then decayed nearly two and a half times faster. By day 110 the adaptive group was ahead by 1,030 steps a day.
The second, ENGAGE, randomised 500 adults at elevated cardiovascular risk — 66 per cent Black, 70 per cent women, from lower-income Philadelphia neighbourhoods — on whether the step goal was self-chosen or assigned. Only the self-chosen arm beat control, by 1,384 steps a day, sustained through follow-up. No arm with an externally assigned goal produced a consistent increase.
Who picks the number matters more than the number.
Does pace matter?
Genuinely unresolved, and anyone telling you otherwise is citing one study and hiding three. In Lee’s 2019 study of 16,741 older women, cadence lost significance once total steps were accounted for (HR 0.86, 0.65–1.13). The CARDIA cohort found the same. But both UK Biobank analyses found the opposite, with peak 30-minute cadence predicting lower risk beyond total volume — most strikingly for dementia, HR 0.38 (0.24–0.60) at 112 steps per minute.
What this evidence cannot tell you
Nobody has randomised step counts and followed people to death. All of this is observational, and the governing threat is reverse causation: undiagnosed illness reduces walking before it causes death. The 2025 synthesis rates its certainty as moderate for most outcomes and low or very low for several.
The plateau may also be partly a measurement artefact. Very few people in these cohorts walk more than about 12,000 steps — in Lee’s study the highest quartile had a median of 8,442 — so confidence intervals widen exactly where the curve appears to flatten. “The benefit levels off” and “we cannot measure it well up there” are hard to tell apart.
Devices are not interchangeable either. A hip-worn research accelerometer, a wrist-worn one and a consumer fitness tracker produce different counts for the same walk, so thresholds from different studies are not measured on the same ruler.
A more useful number
If you already walk a lot, 10,000 is fine and there is evidence behind it. If you do not, the target that matters is your current number plus about a thousand — and the research says you should be the one who picks it.
That is a less satisfying headline than a round figure on a watch face. It is also the only recommendation here that the randomised evidence actually supports.
This article is general information, not medical advice. If you have a heart condition, a mobility problem, or have been inactive for a long time, talk to a doctor before significantly increasing your activity.
