In the most active adults, long sitting carried no measurable extra mortality risk

"Sitting is the new smoking" is a slogan, not a finding. The largest test of the underlying question — a harmonised meta-analysis of more than a million people published in The Lancet — found that among adults in the most physically active quartile, sitting for more than eight hours a day carried a hazard ratio for death of 1.04, with a confidence interval of 0.99 to 1.10 that includes no increase at all [s1]. Among the least active, the association was real and substantial.

That conditionality is the whole story, and it is what the slogan erases. The evidence does not describe sitting as an independent hazard that follows you regardless of what else you do. It describes a risk concentrated in people who are also inactive.

What the million-person analysis found

Researchers pooled individual-level data from 16 studies, 13 of which provided data on sitting time and all-cause mortality, covering 1,005,791 individuals followed for between 2 and 18.1 years, during which 84,609 — 8.4% — died [s1]. Sitting time and TV-viewing time were each sorted into four standardised groups, and physical activity into quartiles measured in MET-hours per week [s1].

Taking as the reference group those who sat under four hours a day and were in the most active quartile (above 35.5 MET-hours per week), mortality rates were 12% to 59% higher across the two lowest activity quartiles [s1]. The extremes are instructive. Sitting more than eight hours a day while doing more than 35.5 MET-hours per week of activity produced a hazard ratio of 1.04 (95% CI 0.99 to 1.10) — statistically indistinguishable from the reference [s1]. Sitting less than four hours a day while in the lowest activity quartile, under 2.5 MET-hours per week, produced a hazard ratio of 1.27 (1.22 to 1.31) [s1].

In other words, in this dataset, being inactive and sitting little was associated with more risk than being active and sitting a great deal.

Television was the exception. Across six studies covering 465,450 people and 43,740 deaths, watching three or more hours a day was associated with increased mortality regardless of physical activity — except in the most active quartile, where the increase appeared only at five hours a day or more (HR 1.16, 1.05 to 1.28) [s1]. The authors concluded that high levels of moderate-intensity activity, about 60 to 75 minutes a day, seem to eliminate the increased risk associated with high sitting time, but only attenuate it for TV viewing [s1].

A second cohort, the same shape

An analysis of 149,077 participants in Australia's 45 and Up Study, published in the Journal of the American College of Cardiology, recorded 8,689 deaths — 1,644 from cardiovascular disease — over a median 8.9 years [s2].

Among participants reporting no moderate-to-vigorous activity at all, the hazard ratio for all-cause mortality comparing the most sedentary group (over eight hours a day) with the least sedentary (under four) was 1.52 (95% CI 1.13 to 2.03) [s2]. Among those meeting either the lower band of the activity recommendation (150 to 299 minutes a week) or the upper band (300 minutes or more), the evidence for elevated risk from sitting was, in the authors' words, inconsistent and weak [s2].

The study also modelled substitution. Among people sitting more than six hours a day, replacing an hour of sitting with vigorous activity was associated with a cardiovascular mortality hazard ratio of 0.36 (0.17 to 0.74) per hour [s2]. Their conclusion: activity doses equivalent to meeting current recommendations attenuate or effectively eliminate the association between sitting and mortality [s2].

What reducing sitting is worth on its own

The most recent estimate comes from an individual participant data meta-analysis published in The Lancet in January 2026, restricted to cohorts with device-measured activity [s3]. It included seven cohorts from Norway, Sweden and the USA (40,327 participants, 4,895 deaths), with UK Biobank data (94,719 participants, 3,487 deaths) analysed separately [s3].

Cutting sedentary time by 30 minutes a day was estimated to prevent 3.0% (95% CI 2.0 to 4.1) of all deaths if targeted at the roughly 20% least active, and 7.3% (4.8 to 9.6) if applied to everyone except the most active [s3]. In UK Biobank the equivalent population-based figure was smaller, at 4.5% (2.8 to 6.1) [s3].

Adding activity looked more valuable than subtracting sitting. A five-minute daily increase in moderate-to-vigorous activity was estimated to prevent 6.0% (4.3 to 7.4) of deaths in the least active group and 10.0% (6.3 to 13.4) in the population-based scenario [s3].

The limits worth stating

The two large cohort analyses rely on self-reported sitting time [s1] [s2], which people estimate badly, and self-report error tends to blur real associations rather than manufacture them. All three studies are observational, so none establishes that sitting causes anything; and reverse causation is a live concern, because people who are ill sit more.

The 2026 analysis is stronger on measurement — it used device-derived activity — but its outputs are modelled potential impact fractions, which are estimates of what might happen under a counterfactual, not observed effects of an intervention [s3].

What survives all of that is the interaction. Sitting time behaves differently depending on how much someone moves, in three separate datasets, using different methods. A slogan that treats it as an unconditional exposure — the way tobacco is an unconditional exposure — is not describing this literature.

This article is informational and is not medical advice.

Sources

• Does physical activity attenuate, or even eliminate, the detrimental association of sitting time with mortality? A harmonised meta-analysis of data from more than 1 million men and women — The Lancet, 2016-07-28
• Sitting Time, Physical Activity, and Risk of Mortality in Adults — Journal of the American College of Cardiology, 2019-04-22
• Deaths potentially averted by small changes in physical activity and sedentary time: an individual participant data meta-analysis of prospective cohort studies — The Lancet, 2026-01-13