What a MET is, and how calories burned is estimated

What a metabolic equivalent actually is

A MET is a ratio, not a unit of energy. It expresses the energy cost of an activity as a multiple of the energy cost of sitting quietly. Walking briskly at 4.3 METs means the activity costs about four and a third times as much energy per minute as resting does. Because it is a ratio, it travels across body sizes in a way a raw calorie figure cannot: 4.3 METs is 4.3 METs whether you weigh 50 kg or 120 kg.

To turn that ratio into oxygen or calories you need a value for the denominator, and the accepted convention sets one MET at 3.5 millilitres of oxygen per kilogram of body mass per minute. Jetté, Sidney and Blümchen's 1990 review is the standard citation for this definition and for the tables of MET values used in exercise testing and prescription that grew out of it. The equivalent expression in energy terms is approximately 1 kilocalorie per kilogram per hour, which is why a 70 kg person is often said to burn about 70 kilocalories an hour doing nothing.

The awkward history is that the 3.5 figure was not derived from a population. Byrne and colleagues traced it to the measured resting oxygen consumption of a single individual: a 70 kg, 40-year-old man. It became a convention because it was convenient and because nothing better was proposed, and it has been embedded in exercise physiology ever since. It is a fine reference point. It is not a description of you.

The Compendium of Physical Activities, and how its values were derived

The MET value for any given activity comes from the Compendium of Physical Activities, first published by Ainsworth and colleagues in 1993 to give researchers a shared coding system for physical activity questionnaires. It has been updated three times since: a second edition in 2000, a major update in 2011, and the 2024 Adult Compendium, which now sits alongside separate compendia for older adults and for wheelchair users.

The values were assembled from published studies that measured oxygen consumption during specific activities, usually by indirect calorimetry in small groups of adults. Where good measurements existed, the compendium reports the measured average. Where they did not (and for a substantial share of the more than a thousand listed activities they did not), the value was estimated by analogy with a similar activity or from the physiological demands of the movement. The compendium has always been explicit about this, and the current tables carry a code indicating the origin of each value.

Two limitations follow directly from how it was built. The first is that the samples behind the measured values were typically small, often young, and frequently healthier than average, so the numbers describe a narrow slice of humanity. The second is that a single MET value has to cover an enormous range of real-world execution: cycling at moderate effort is 8.0 METs whether you are grinding up a rise or freewheeling behind someone, and weight training at vigorous effort is 6.0 METs whether your rest periods are forty seconds or four minutes.

None of this makes the compendium a bad instrument. It is the best available shared vocabulary for the energy cost of activity, it is maintained openly, and epidemiology built on it has produced most of what we know about physical activity and health. It just answers a question about populations, and people keep asking it a question about individuals.

The arithmetic, written out

Once you have a MET value the calculation is short. Multiply the MET value by 3.5 and by your body mass in kilograms, divide by 200 to convert millilitres of oxygen into kilocalories per minute, then multiply by the duration in minutes. The division by 200 embeds the standard assumption that one litre of oxygen consumed yields about five kilocalories.

Worked through for a 70 kg adult walking briskly at 5.6 km/h (4.3 METs on the compendium) for 30 minutes: 4.3 × 3.5 × 70 ÷ 200 gives 5.27 kilocalories per minute, and over 30 minutes that is 158 kilocalories. The same walk for a 100 kg adult comes to 226 kilocalories, because the whole calculation scales linearly with body mass.

That linear scaling is both the method's greatest strength and a quiet source of error. It is right in the sense that moving a heavier body the same distance genuinely costs more energy. It is wrong in the sense that it assumes resting metabolism per kilogram is constant across body sizes, when in fact a heavier body typically carries a larger proportion of fat mass, which is metabolically much less active per kilogram than lean tissue. The estimate therefore tends to be generous to larger people and to anyone carrying more fat than average.

Why a MET figure is a population average with wide scatter

Byrne and colleagues put the convention to the test directly. They measured resting metabolic rate by indirect calorimetry in 769 adults: 642 women and 127 men, aged 18 to 74, weighing between 35 and 186 kilograms. Average resting oxygen consumption came out at 2.6 millilitres per kilogram per minute, with a standard deviation of 0.4, against the conventional value of 3.5. Expressed in energy terms, average resting expenditure was 0.84 kilocalories per kilogram per hour against the conventional 1.0.

Their conclusion is worth stating precisely: for a large, heterogeneous sample the 1-MET value of 3.5 ml per kilogram per minute overestimates actual resting oxygen consumption by about 35% on average, and the 1 kilocalorie per kilogram per hour figure overestimates resting energy expenditure by about 20%. That is a systematic bias, not random noise, and it runs in the direction of flattery.

The same study explains where the individual variation comes from. Body composition (fat mass and fat-free mass together) accounted for 62% of the variance in resting oxygen consumption per kilogram, while age accounted for only 14%. Fat-free mass is metabolically expensive and fat mass is comparatively cheap, so two people of identical weight and age can have genuinely different resting rates depending on what that weight is made of. Any MET-based calorie figure inherits all of that variation and shows you none of it.

What this means in practice is modest and specific. A MET-derived figure is a reasonable central estimate for a group and a rough one for a person, and it will more often be too high than too low. It remains perfectly good for comparing activities against each other, for tracking your own sessions over time, and for checking a week's activity against public health guidance in MET-minutes, all comparisons in which a consistent bias largely cancels out.

Why wearables and cardio machines disagree with each other and with the truth

A wrist device does not measure energy expenditure. It measures movement with an accelerometer, usually heart rate with an optical sensor, and then applies a proprietary model to guess at calories. Shcherbina and colleagues tested six commercial wrist-worn devices against indirect calorimetry in 60 participants across walking, running and cycling. Heart rate measurement held up reasonably well. Energy expenditure did not: no device tested reached an acceptable error rate, and the best of them still missed by margins that make a daily calorie budget unusable.

The reason is structural rather than a failure of any one manufacturer. Accelerometer counts have a genuinely useful but imperfect relationship with metabolic cost (Kozey and colleagues showed how much that relationship varies between activities), and heart rate is confounded by heat, caffeine, stress, dehydration and cardiovascular drift, all of which raise the pulse without raising the workload. A model fed two noisy inputs and asked to produce a precise output will produce a precise-looking output, and the precision is manufactured.

Cardio machines have a different set of problems. A treadmill knows speed and gradient, which are genuinely informative, but unless you have entered your weight it assumes a default (frequently 70 or 75 kilograms), and the entire calculation scales linearly with that number. An elliptical or a stair machine has no reliable way to know how much of the work you are actually doing rather than letting momentum do, and most consoles report the gross figure, which quietly includes the calories you would have burned sitting in the café instead.

The consistent finding across device comparisons is that the errors are not only large but inconsistent between devices, so two people doing the same session with different watches will be told different things, and neither will match a metabolic cart. Use the number as a relative index of your own sessions, where a consistent bias subtracts itself out, and never as an input to a food budget.

Gross against net, and why the difference matters

There are two honest answers to how many calories an activity burned. The gross figure is the total energy your body used during the session. The net figure subtracts the energy you would have used anyway, lying on the sofa for the same length of time, and so represents the additional cost of having done the activity. Net is the number that matters if you are thinking about energy balance, because the resting portion is already counted inside your total daily energy expenditure.

Arithmetically the difference is one MET's worth of time: net calories use (MET − 1) in place of MET. That sounds trivial and is not. At low intensities the subtracted MET is a large share of the total, so an hour of Hatha yoga at 2.5 METs costs a 70 kg adult 184 kilocalories gross but only 110 net. At high intensities the share is small: half an hour of running at 9.8 METs costs 412 gross and 370 net.

Almost every consumer device and machine reports the gross figure, because it is the larger one. When you then add that number to your day and subtract it from your food, you have double-counted your resting metabolism for the duration of the session. Over an hour of light activity every day that is around 70 kilocalories of phantom deficit, which is enough to explain a stalled month.

Why exercise burn is smaller than most people expect

Set the numbers above against food and the scale of the problem becomes obvious. A brisk half-hour walk nets a 70 kg adult about 121 kilocalories, which is roughly a small flat white and a biscuit. A hard 45-minute cycle nets around 386, which is a little under a large muffin. This is not an argument against exercising (the cardiovascular, metabolic, musculoskeletal and mental health returns on training are among the best established findings in medicine), but it is a decisive argument against using exercise as your primary lever on body weight.

There is a second effect that makes the gap wider still. Pontzer and colleagues, studying total daily energy expenditure with doubly labelled water across a wide range of activity levels, found that expenditure does not rise in simple proportion to activity. Beyond a moderate level, additional physical activity was associated with much smaller increases in total daily expenditure than a straight addition would predict, apparently because the body economises elsewhere: in resting metabolism, in spontaneous movement, or in other physiological processes. The session's calories are real; they just do not all survive to the end of the day.

The practical implications are straightforward and slightly unwelcome. Do not eat back the calories a device reports, because the report is gross rather than net, is biased high by the 3.5 convention, and is being partly offset by compensation you cannot see. Judge a training programme by what it does to your performance, your strength, your resting heart rate and your waist measurement rather than by the calorie counter. And if a weight goal is the objective, set the deficit on the intake side where the numbers are large and measurable, using a total daily energy expenditure estimate as the starting point and your own weight trend as the correction.

Frequently asked questions

Put it into practice

Run your own numbers through the calories burned calculator, the TDEE calculator and the BMR calculator. Related reading: TDEE explained, Heart rate zones and Estimating a 1RM.

Sources

1. Jetté M, Sidney K, Blümchen G. Metabolic equivalents (METS) in exercise testing, exercise prescription, and evaluation of functional capacity. Clin Cardiol 1990;13:555–65. doi.org/10.1002/clc.4960130809
2. Byrne NM, Hills AP, Hunter GR, Weinsier RL, Schutz Y. Metabolic equivalent: one size does not fit all. J Appl Physiol 2005;99:1112–9. doi.org/10.1152/japplphysiol.00023.2004
3. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of Physical Activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 1993;25:71–80. doi.org/10.1249/00005768-199301000-00011
4. Ainsworth BE, Haskell WL, Herrmann SD, et al. 2011 Compendium of Physical Activities: a second update of codes and MET values. Med Sci Sports Exerc 2011;43:1575–81. doi.org/10.1249/MSS.0b013e31821ece12
5. Herrmann SD, Willis EA, Ainsworth BE, et al. 2024 Adult Compendium of Physical Activities: a third update of the energy costs of human activities. J Sport Health Sci 2024;13:6–12. doi.org/10.1016/j.jshs.2023.10.010
6. Compendium of Physical Activities: the maintained 2024 Adult Compendium MET value tables. pacompendium.com/adult-compendium/
7. Kozey SL, Lyden K, Howe CA, Staudenmayer JW, Freedson PS. Accelerometer output and MET values of common physical activities. Med Sci Sports Exerc 2010;42:1776–84. doi.org/10.1249/MSS.0b013e3181d479f2
8. Shcherbina A, Mattsson CM, Waggott D, et al. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med 2017;7:3. doi.org/10.3390/jpm7020003
9. Pontzer H, Durazo-Arvizu R, Dugas LR, et al. Constrained total energy expenditure and metabolic adaptation to physical activity in adult humans. Curr Biol 2016;26:410–17. doi.org/10.1016/j.cub.2015.12.046