Around one calorie per kilogram of body weight per kilometre. That rule of thumb is roughly right, and almost everything about how it is used is wrong — starting with the fact that it was derived from two people.
The rule, and where it came from
The figure everyone quotes — 1 kcal per kg of body weight per km — comes from a 1963 paper. The sample was two subjects, both Olympic middle-distance runners.
The authors were careful in ways that later users were not. They wrote that the net cost “is approximately 1 kcal/kg km, running on the level”, stated that values “must be corrected by ±3%, depending on whether an untrained subject or a trained athlete is used”, and excluded wind resistance by design.
Two things in that sentence matter. It is approximate, and it is net — resting metabolism was subtracted out.
Later work broadly supports the central estimate. Thirty-six amateur runners came out at about 0.90 kcal/kg/km. Twenty runners of mixed ability averaged 0.98 and 1.02. So the constant is sound as a group average.
The individual spread is where it breaks. In that last study, measured values ranged from 0.85 to 1.25 kcal/kg/km — the least economical runner burned 47% more per kilometre than the most economical one. That 47% is the gap between the two extremes of a 20-person sample, not a typical spread; the figure usually quoted for variation in running economy among trained runners is around 30%. Either way, a single formula cannot be right for everyone.
So for a 70 kg runner covering 10 km: about 700 kcal on the rule, but plausibly anywhere from 600 to 875 depending on who you are. Nobody knows which you are without measuring you.
Net versus gross: smaller than you would think
You were going to burn calories during that hour whether you ran or sat still. The honest figure subtracts that.
Working from published metabolic equivalents for running, the resting share of a gross running-calorie figure is about 7% at fast paces and 12% at slow ones — roughly 35 kcal on a 30-minute run for a 70 kg runner, or 70 kcal over an hour.
Note what it depends on. Not duration — both terms scale with time — but intensity. The slower you go, the larger the share of the total you would have burned anyway.
The bigger problem is that the constant underneath all of this is itself wrong. The value defining 1 MET “was first derived from the resting O2 consumption of one person, a 70-kg, 40-yr-old man”. Measured against 769 real adults, it overestimates resting energy expenditure by 20%.
Two of the foundational constants in every running-calorie calculator were derived from one person and two people respectively. That is worth sitting with before trusting a number to three digits.
Your watch is not measuring this
This is the practical part, and the numbers are worse than most people assume.
Stanford researchers tested seven wrist devices against clinical-grade indirect calorimetry in 60 people spanning a wide range of body sizes and skin tones. For the running task, the best relative error across devices was 31.0% (95% CI 28.0–34.0). Their conclusion was blunt:
“none of the devices provided estimates of energy expenditure that were within an acceptable range in any setting.”
The same devices measured heart rate to within about 5%. The hardware is fine. Turning heart rate into calories is the part that does not work.
Two systematic reviews reach the same place. One, pooling 104 comparisons, found devices underestimated energy expenditure overall and said the authors “would be hesitant to consider any device sufficiently accurate”, with the best device still ranging from −21% to +15%. Another, covering 158 publications and 5,934 participants, found that only 9.2% of controlled-setting energy comparisons fell within ±3% — and that no brand met acceptable accuracy for energy expenditure, while all brands were acceptable for heart rate.
Running-specific tests give the same picture. In twenty trained runners on a treadmill, errors at aerobic intensities ran from 10% to 42% depending on device, deteriorating sharply at high intensity. Outdoors at 10 km/h, three popular watches came in at 24.4%, 21.8% and 11.9% error.
On gym machines, we are not going to give you a number. The widely circulated figures — treadmill 13% over, elliptical 42% over — trace to a 2010 laboratory demonstration publicised through news coverage, with no peer-reviewed publication, no stated sample size and no published methods. We could not find a primary source, so we will not repeat it. What we can say is that the wearable evidence above is real, and there is no good reason to think a treadmill console does better.
The afterburn is not what you were told
EPOC — the elevated metabolism after exercise — is real. Its size is small.
The definitive review found that prolonged EPOC requires a serious stimulus (at least 50 minutes at 70% of VO2max, or supramaximal intervals), and that even then the EPOC comprises only 6–15% of the net oxygen cost of the exercise. Its verdict:
“the earlier research optimism regarding an important role for the EPOC in weight loss is generally unfounded… The role of exercise in the maintenance of body mass is therefore predominantly mediated via the cumulative effect of the energy expenditure during the actual exercise.”
For a 70 kg runner, that works out as:
- 30 min at 9 km/h: ~315 kcal in the run, ~19–47 kcal afterwards
- 45 min at 10 km/h: ~525 kcal, ~32–79 kcal afterwards
- 60 min at 11 km/h: ~770 kcal, ~46–116 kcal afterwards
Twenty to fifty calories for a typical club run — less than the measurement error of the watch reporting it. And most easy runs will not even reach the intensity threshold required to produce a prolonged EPOC at all.
Two things that do change the number
Body mass, and it scales cleanly. When researchers compared severely obese runners (108 to 172 kg) with normal-weight runners at 8 km/h, mass-specific cost was independent of body mass — meaning cost per kilogram is roughly constant, so total cost scales with how much you weigh. A heavier runner genuinely burns proportionally more over the same distance.
One caution: that is different from strapping load onto a given person. Adding 10% of body mass and weight raised net metabolic cost by 14 ± 2.7%, 20% raised it by 24 ± 3.2%, and 30% by 38 ± 3.8% — more than proportionally.
The clever part of that study is what it separates. When the researchers added the mass without the weight, using a harness to offset the load, the cost barely moved: about 5% at 110% of body mass, and no significant difference at all at 120% or 130%. So it is not carrying extra matter that costs you, it is generating the force to support extra weight — which the authors put at up to 74% of the net cost of running. A habitually heavier body has adapted. A weighted vest has not. Do not use the second finding to make claims about the first.
Speed, but only outside a narrow band. The claim that cost per kilometre is speed-independent holds “for a narrow range of moderate running speeds (2.68–3.58 m·s−1), but not slower or faster.” That is roughly 9.6 to 12.9 km/h. Outside that band the cost per kilometre goes up in both directions: at a slow jog of about 6.4 km/h it was around 14% higher than at 9.6 km/h, and at 18.5 km/h it was roughly 10 to 13% higher than across the moderate range. Very slow running is not cheaper per kilometre — it is dearer. (Those figures are from sub-elite runners on a treadmill; the shape holds more generally.)
Outdoors, add air resistance — about 8% of total energy cost at 21.5 km/h, and negligible at club paces.
The practical version
- Estimate: your weight in kg × distance in km ≈ calories. For a 70 kg runner, a 5K is about 350 kcal and a half marathon about 1,500.
- Then treat it as ±25%, because of individual economy alone.
- Subtract about 10% if you want the honest net figure rather than the gross one.
- Do not add anything for afterburn. On a normal run it is 20 to 50 calories and probably zero.
- Do not eat back what the watch says. A 30% overestimate on a 500 kcal run is 150 calories that were never burned — enough, repeated, to erase the run entirely.
The useful conclusion is not a better number. It is that running-calorie figures are far too imprecise to manage a diet by. If food is what you want to change, measure the food, not the run.
References
- Margaria R, et al. Energy cost of running. J Appl Physiol, 1963;18(2):367–70. doi:10.1152/jappl.1963.18.2.367
- di Prampero PE, et al. The energetics of endurance running. Eur J Appl Physiol, 1986;55(3):259–66.
- Batliner ME, et al. Does metabolic rate increase linearly with running speed in all distance runners? Sports Med Int Open, 2018;2(1):E1–E8. doi:10.1055/s-0043-122068
- Barnes KR, Kilding AE. Running economy: measurement, norms, and determining factors. Sports Med Open, 2015;1:8. doi:10.1186/s40798-015-0007-y
- Byrne NM, et al. Metabolic equivalent: one size does not fit all. J Appl Physiol, 2005;99(3):1112–9. doi:10.1152/japplphysiol.00023.2004
- Shcherbina A, et al. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med, 2017;7(2):3. doi:10.3390/jpm7020003
- O’Driscoll R, et al. How well do activity monitors estimate energy expenditure? Br J Sports Med, 2020;54(6):332–40. doi:10.1136/bjsports-2018-099643
- Fuller D, et al. Reliability and validity of commercially available wearable devices. JMIR Mhealth Uhealth, 2020;8(9):e18694. doi:10.2196/18694
- LaForgia J, Withers RT, Gore CJ. Effects of exercise intensity and duration on the excess post-exercise oxygen consumption. J Sports Sci, 2006;24(12):1247–64. doi:10.1080/02640410600552064
- Taboga P, et al. Energetics and mechanics of running men: the influence of body mass. Eur J Appl Physiol, 2012;112(11):4027–33. doi:10.1007/s00421-012-2389-6
- Teunissen LPJ, Grabowski A, Kram R. Effects of independently altering body weight and body mass on the metabolic cost of running. J Exp Biol, 2007;210(24):4418–27. doi:10.1242/jeb.004481
- Pugh LGCE. Oxygen intake in track and treadmill running with observations on the effect of air resistance. J Physiol, 1970;207(3):823–35.