If you want the current best answer to how much exercise to offset sitting all day, it is roughly 30 to 40 minutes of moderate-to-vigorous activity per day. That figure comes from a harmonised meta-analysis of nine cohorts and 44,370 people whose activity was measured with accelerometers, and the authors chose their verb carefully: that amount of activity attenuates the association between sedentary time and death. Attenuate, not eliminate. The more familiar “about an hour a day” number came from an earlier analysis built on what people said they did, and the 2020 paper states plainly that the accelerometer-based estimate is “lower than previous estimates from self-reported data”.
Nobody flip-flopped. The instrument changed.
Picture the position this puts a working engineer in. Suppose you are on call, you sit nine or ten hours on an ordinary day, and everything you have read for a decade says the price is an hour of exercise, daily, forever. You cannot buy that hour. So you file the subject under true but not actionable, which is the most expensive conclusion available, because the figure you were failing to hit was never the figure the device-measured evidence points at.
How much exercise to offset sitting all day: where the hour came from
In 2016, Ekelund and colleagues pooled data from more than a million people in the Lancet and reported that “high levels of moderate intensity physical activity (ie, about 60-75 min per day) seem to eliminate the increased risk of death associated with high sitting time”. That is the sentence that became the hour. The same sentence carries a limit that rarely travels with it: the high activity level “attenuates, but does not eliminate” the increased risk associated with high TV-viewing time.
Both the sitting and the activity in that analysis were self-reported, estimated by participants and harmonised across sixteen studies. Note also the authors’ hedge, “seem to eliminate”, describing a statistical association in observational data rather than a demonstrated cancellation of harm.
The comparison buried inside that paper is worth keeping. Against a referent group who sat under four hours a day and fell in the highest activity quartile, people who sat more than eight hours a day but reported more than 35.5 MET-hours per week showed a hazard ratio of 1.04 (95% CI 0.99 to 1.10). Those who sat the least, under four hours a day, but were in the lowest activity quartile at under 2.5 MET-hours per week, came in at 1.27 (95% CI 1.22 to 1.31). Set those two numbers side by side and, in that dataset, low activity tracked with a larger increase in mortality risk than long sitting did. Both rest on self-reported exposures, and both describe associations, not causes.
What happened when the same author used devices
Four years later, in the British Journal of Sports Medicine, the same lead author ran the same kind of harmonised analysis on accelerometer data: nine prospective cohorts from four countries, 44,370 men and women, followed 4.0 to 14.5 years, 3,451 deaths, a 7.8% mortality rate. The conclusion was that higher sedentary time is associated with higher mortality in less active individuals when measured by accelerometry, and that “about 30-40 min of MVPA per day attenuate the association between sedentary time and risk of death, which is lower than previous estimates from self-reported data”.
That final clause is the whole story. The 2016 figure was not retracted and has not been declared wrong. It was superseded on better measurement, by the same lead author, on the same question.
Why would self-report inflate the requirement? This next part is my reasoning rather than a finding in the packet: people are poor instruments for their own behaviour, and the errors do not cancel. If reported activity is noisy relative to what a device records, the estimated dose needed to shift an outcome drifts upward, because the reported hour stands in for something less than an hour of movement. The papers do not spell out that mechanism. They report the discrepancy and its direction.
The cohorts in that 2020 analysis were middle-aged and older adults. That matters, and it keeps mattering below.
Sitting still carries an association after activity is accounted for
It would be convenient if exercise made sitting irrelevant. The dose-response evidence says otherwise, though the effect per hour is small. A meta-analysis of thirty-four studies covering 1,331,468 unique participants, adjusted for physical activity, found a non-linear relationship for all-cause mortality: a relative risk of 1.01 (1.00 to 1.01) per additional hour per day at or below eight hours, rising to 1.04 (1.03 to 1.05) per hour above eight. For all-cause and cardiovascular mortality, the authors identified a threshold of 6 to 8 hours a day of total sitting, and 3 to 4 hours of TV viewing, above which risk is increased.
Those per-hour numbers are small, and reporting them as anything else would be dishonest. What they set is the shape: the curve bends, and it bends near where a normal engineering workday already sits.
The shape shows up again in a 2019 accelerometry-based pooled analysis of eight studies, n=36,383, median follow-up 5.8 years. Hazard ratios across quarters of sedentary time ran 1.00 for the least sedentary, then 1.28 (1.09 to 1.51), 1.71 (1.36 to 2.15) and 2.63 (1.94 to 3.56). Its conclusion: higher total physical activity at any intensity, and less sedentary time, are associated with substantially reduced risk of premature mortality, non-linearly. The scope statement is the authors’ own, middle-aged and older adults, and the sample had a mean age of 62.6 years and was 72.8% women. If you are a 34-year-old engineer, those hazard ratios are not your numbers, and nothing in the paper claims they transfer.
What “offset” does and does not mean
Every study above has one kind of outcome: death. All-cause mortality throughout, and cardiovascular mortality in the sitting threshold analysis. None of them measured your lower back, your neck, your eyes, your afternoon concentration or your mood. When a headline says exercise offsets sitting, the claim on the table is mortality in a large observational cohort, nothing else. If your complaint about your desk is a stiff back at 16:00, this literature does not speak to it.
The second boundary is causal. These are cohort studies. They report that people who moved more died less often over follow-up, adjusted for what the researchers could adjust for. They do not establish that adding activity to your week will move your personal risk, and no honest reading turns a hazard ratio into a number about you.
Separately from the offset question, the WHO puts the public-health floor at “at least 150 minutes of moderate-intensity physical activity per week”, and states that insufficiently active people have a 20% to 30% increased risk of death compared with sufficiently active people. Do not fold that into the 30 to 40 minute figure. They answer different questions: the WHO recommendation is a general population minimum, the Ekelund figure is specifically about the association between high sedentary time and mortality.
Starting or increasing exercise when you have an existing condition, a recent injury or symptoms that worry you is a question for a clinician who can examine you, not for a meta-analysis and not for this site. See our medical disclaimer.
Common questions
Is half an hour a day enough?
Half an hour sits inside the range the device-measured evidence points at. A harmonised meta-analysis of nine cohorts and 44,370 people using accelerometers concluded that about 30 to 40 minutes of moderate-to-vigorous physical activity per day attenuates the association between sedentary time and risk of death, lower than previous estimates from self-reported data. Two caveats travel with that: the studies are observational, so this is association and not demonstrated cause, and the cohorts were middle-aged and older adults.
Was the “one hour a day” figure wrong?
It was not retracted and has not been shown to be wrong. The 2016 Lancet meta-analysis of self-reported data from more than a million people found that about 60 to 75 minutes a day of moderate activity seemed to eliminate the increased mortality risk associated with high sitting time. The same lead author later repeated the approach using accelerometers and arrived at 30 to 40 minutes, which that paper describes as lower than previous estimates from self-reported data. The right description is superseded on better measurement, not overturned.
Does very short, intense movement count for anything?
It is associated with something, in a specific group. A UK Biobank analysis of wearable-measured vigorous intermittent lifestyle physical activity in 25,241 non-exercisers, mean age 61.8 years, followed an average of 6.9 years with 852 deaths, found the sample median of 4.4 minutes a day was associated with a 26% to 30% reduction in all-cause and cancer mortality risk and a 32% to 34% reduction in cardiovascular mortality risk. Those are relative reductions in a self-declared non-exercising, older group whose comparison point was no vigorous activity at all. They are not absolute risk changes, the analysis is observational, and 4.4 minutes is not a demonstrated replacement for the 30 to 40 minute figure.
If I hit the activity target, can I stop worrying about hours in the chair?
The evidence does not support that. A meta-analysis of thirty-four studies covering 1,331,468 participants, adjusted for physical activity, still found a threshold of 6 to 8 hours a day of total sitting above which all-cause and cardiovascular mortality risk is increased, with an all-cause relative risk per additional hour of 1.01 (1.00 to 1.01) below 8 hours a day and 1.04 (1.03 to 1.05) above it. Those per-hour figures are small, and the studies are observational, but the association survives adjustment for activity rather than disappearing.
Do these findings apply to someone in their thirties?
Not directly, and the papers say so. The 2019 pooled accelerometry analysis had a mean age of 62.6 years and was 72.8% women, and states its own scope as middle-aged and older adults; the vigorous-activity analysis had a mean age of 61.8 years. The general direction, that more total activity at any intensity and less sedentary time are associated with lower mortality risk, is what the evidence supports. The specific hazard ratios belong to those older populations, not to a younger reader.
The correction matters less for its size than for what it exposes about the framing. “Offsetting” implies a ledger, hours in the chair on one side, minutes of penance on the other, a balance you either clear or fail to clear. The device-measured evidence describes no ledger. It describes two associations that both bend: more total activity at any intensity tracks with lower mortality risk, more sedentary time tracks with higher, and the steep part of the sedentary curve starts past six to eight hours. Neither cancels the other.
That is a better position than the hour implied, because a ledger you cannot clear is a reason to quit, and a curve is a reason to move along it. How you break up the sitting itself is a separate question with its own literature, covered in how often you should get up from sitting. If you want to know how we read and check papers like these before writing about them, see how we research.
Leave a Reply