The most-measured aerobic dose there is
You can change this. Whether hitting a daily target itself cuts death is untested — no trial has randomized adults to a step target and measured all-cause mortality.
Roughly 7,000–8,000 steps a day, with the plateau arriving earlier after 60. Every number supporting it is observational.
Dose
7,000–8,000 steps/day under 60; 6,000–8,000 from 60 on. Accumulated any way across the day — the 10-minute minimum bout was removed from the US guidelines in 2018.
Dose-response shape: plateau
How to get there
Walk however the day allows — 7,000–8,000 steps under 60, 6,000–8,000 from 60 on. No minimum bout length is required.
Steps per day, tracked over weeks. Inoue et al. 2023 found hitting the target on just 1–2 days a week already carried most of the benefit versus zero days, plateauing at 3–7 days.
Banach et al. 2023 pooled a linear 'more is better' relationship with no inflection point — against the plateau this target assumes. The shape past the target is not settled.
Evidence, by endpoint
| Endpoint | Grade | Citation |
|---|---|---|
| All-cause mortality | B | Ding et al. 2025, The Lancet Public Health 14-study dose-response pooling. 7,000 vs 2,000 steps/day HR 0.53 (95% CI 0.46–0.60), I² 36.3%, with the curve inflecting around 5,000–7,000 steps. |
| All-cause mortality, by age band | B | Paluch et al. 2022, The Lancet Public Health Harmonised individual-participant meta of 15 cohorts (47,471 adults, 3,013 deaths). The plateau sits at 6,000–8,000 steps/day at 60 and over, and 8,000–10,000 under 60 (p for interaction 0.012). |
| All-cause mortality per 1,000-step increment | B | Banach et al. 2023, European Journal of Preventive Cardiology 17 cohort studies, 226,889 people. Each additional 1,000 steps/day HR 0.85 (0.81–0.91) — pooled as a linear increment from about 3,867 steps upward, a different curve shape from the engine paper rather than a confirmation of it. |
| Mortality and cardiovascular events from sustained bouts vs sporadic bursts | C | Del Pozo Cruz et al. 2025, Annals of Internal Medicine Among suboptimally active adults, walking in bouts of ≥15 minutes tracked with lower 9.5-year mortality (0.80% vs 4.36%) and CVD incidence (4.39% vs 13.03%) independent of total step volume. Absolute risks in a single cohort, not a pooled hazard ratio. |
| Step intensity (cadence) independent of total volume | X | Saint-Maurice et al. 2020, JAMA NHANES cohort of 4,840 adults. Volume tracked strongly with mortality (8,000 vs 4,000 steps/day HR 0.49; 12,000 vs 4,000 HR 0.35), but step intensity showed no independent association once total volume was adjusted for. Walking faster is not the lever here; walking more is. |
| All-cause mortality from a randomized step-count intervention | X | No trial has randomized adults to a daily step target and measured all-cause mortality. |
What the dose is
Seven to eight thousand steps a day for adults under 60, and six to eight thousand from 60 on, accumulated however the day allows. The 2018 US Physical Activity Guidelines removed the previous 10-minute minimum bout duration, so there is no requirement that the volume arrive in blocks.
What the evidence shows
Ding et al. 2025 (The Lancet Public Health) pooled a 14-study dose-response and found 7,000 versus 2,000 steps/day at HR 0.53 (95% CI 0.46–0.60), with the curve inflecting around 5,000–7,000 steps. Paluch et al. 2022 (The Lancet Public Health), a harmonised individual-participant meta of 15 cohorts, puts the plateau at 6,000–8,000 steps/day for adults 60 and over and 8,000–10,000 for those under 60 — the age split Ding lists as a limitation of its own paper. Both are meta-analyses of prospective cohorts with a dose-response, which is grade B here and not grade A.
What the number is not
It is not an intensity prescription. Saint-Maurice et al. 2020 (JAMA) found step cadence had no independent association with mortality once total daily volume was accounted for, so the lever is how much you walk rather than how fast. It is also not a randomized result: no trial has assigned adults to a step target and measured all-cause mortality, which is why the mortality claim tops out at grade B no matter how many cohorts agree.
How the volume gets accumulated
Del Pozo Cruz et al. 2025 (Annals of Internal Medicine) found that among suboptimally active adults, steps taken in sustained bouts of 15 minutes or more tracked with lower 9.5-year mortality (0.80% vs 4.36%) and lower CVD incidence (4.39% vs 13.03%) independent of total step volume. That is a single cohort reporting absolute risks, so it is graded C — enough to prefer a deliberate walk over the same steps scattered across errands, not enough to require it.
What argues against this
Every figure above is observational, and the reference group is doing much of the work. Ding's headline HR 0.53 compares 7,000 steps against 2,000 — and that paper's own note is that 2,000 is the lower bound of the normal range for older adults, not a typical adult, so a large share of the contrast is the gap between a walking adult and a barely mobile one. Reverse causation runs the same direction: illness lowers step counts long before it kills. The shape is not settled either — Banach et al. 2023 pools a linear 'more is better' from about 3,867 steps/day upward with no inflection, directly against the 5,000–7,000 plateau this target is built on. And Inoue et al. 2023 (JAMA Network Open) found that hitting 8,000 steps on just 1–2 days a week carried an adjusted 10-year mortality risk difference of −14.9 percentage points (95% CI −18.8 to −10.9) versus zero such days, plateauing by 3–7 days — evidence about pattern rather than daily average, and an argument against prescribing a daily number at all.
Gotchas
- The 2,000-step reference in the headline comparison is not an ordinary sedentary adult — it is the lower bound of the normal range for older adults. Reading 'halves your mortality risk' off that contrast overstates what moving from a normal baseline buys.
- Ding et al. 2025 do not treat 7,000 as a ceiling: 10,000 remains viable for more active people, and 7,000 is the more realistic target for some. The plateau is where the curve bends, not where benefit stops.
- These cohorts measure steps with worn research accelerometers. A phone carried for part of the day is counting a different quantity.
- Del Pozo Cruz et al. 2022 (JAMA Internal Medicine, 78,500 UK Biobank adults) reports its dose-response as change in log-HR per 2,000-step increment, not as fixed percentage reductions. Percentages quoted off that paper elsewhere are someone else’s arithmetic.
Educational content only — not medical advice, diagnosis, or treatment. Not a medical device; not FDA evaluated. Consult a qualified healthcare professional about your own health, and call emergency services for urgent symptoms. Full medical disclaimer →