Calories Burned by Heart Rate: How to Work It Out and How Far Off a Watch Can Be
Calories burned by heart rate come from the Keytel formula: pulse, weight and age in, energy out. A man of 38 at 150 beats per minute for 40 minutes gets 607 kcal, within a range of 366 to 728.
Key takeaways
- The Keytel formula turns average heart rate, weight and age into energy: each beat per minute adds 0.6309 kJ for men and 0.4472 kJ for women, and dividing by 4.184 gives kilocalories.
- A man of 38 at 82 kg and 150 bpm for 40 minutes comes to 607 kcal; a woman of 29 at 62 kg and 145 bpm for 30 minutes comes to 278 kcal.
- In the paper's check on 17 new exercisers, estimates ran 3.0 kcal a minute too low to 6.0 too high (95% limits), and 1.5 kcal a minute high on average.
- No wrist device in a 60-person test kept its energy error under 20%; median errors ran from 27.4% to 92.6%, while the Apple Watch's pulse error was 2.0%.
- The three sample sessions came out 5% to 19% above the matching Compendium MET figure.
Calories burned by heart rate come from a 2005 equation, the Keytel formula, which turns your average pulse, weight and age into energy per minute. Men and women get separate versions. Put in a man of 38 who weighs 82 kg and holds 150 beats per minute for 40 minutes, and the answer is 607 kcal. The real figure could sit anywhere from 366 to 728 kcal. A wrist watch can miss by more.
How do you work out calories burned by heart rate?
Use the Keytel equation for your sex, add up four terms, then divide by 4.184 to turn kilojoules into kilocalories. Keytel and colleagues built it from 115 regular exercisers aged 18 to 45 and weighing 47 to 120 kg. Each person did three steady stages on a treadmill or a bike, at about 57%, 77% and 90% of maximum heart rate. Gas analyzers measured the oxygen they used and the carbon dioxide they breathed out. That gave the reference burn. The authors fitted two models. One needs a lab fitness score (VO2 max). The other doesn’t, and I use it, because almost nobody has that score.
Here are its coefficients. They give kilojoules per minute.
| Term | Men | Women |
|---|---|---|
| Constant | −55.0969 | −20.4022 |
| Each beat per minute | 0.6309 | 0.4472 |
| Each kilogram of weight | 0.1988 | −0.1263 |
| Each year of age | 0.2017 | 0.074 |
I read these numbers in the paper. I also matched them against two open-access papers that reprint them, Jung and colleagues (2021) and Chiu and colleagues (2022).
Walk through the man of 38 at 82 kg and 150 beats per minute. The pulse term is 0.6309 × 150 = 94.635. The weight term is 0.1988 × 82 = 16.302, and the age term is 0.2017 × 38 = 7.665. Add the constant of −55.0969 and you get 63.50 kJ a minute. Divide by 4.184 and that is 15.18 kcal a minute. Forty minutes makes 607 kcal.
Two more sessions use the same steps. The woman of 29 at 62 kg and 145 beats per minute gets 64.844 from pulse and 2.146 from age. Weight takes away 7.831, and the constant starts her at −20.4022. That comes to 38.76 kJ a minute, or 9.26 kcal. The woman of 44 at 78 kg and 125 beats per minute lands at 28.90 kJ a minute, or 6.91 kcal.
| Person and session | Average pulse | kJ a minute | kcal a minute | Session total |
|---|---|---|---|---|
| Man, 38, 82 kg, run, 40 min | 150 bpm | 63.50 | 15.18 | 607 kcal |
| Woman, 29, 62 kg, run, 30 min | 145 bpm | 38.76 | 9.26 | 278 kcal |
| Woman, 44, 78 kg, brisk walk, 60 min | 125 bpm | 28.90 | 6.91 | 415 kcal |
I chose these sessions to differ in pulse, age, weight and length. They aren’t anyone’s measurements.
Two features of the equations matter before you trust them. Each extra beat per minute adds 0.15 kcal a minute for men and 0.11 for women (0.6309 and 0.4472, divided by 4.184). Ten beats of difference is 60 kcal over the man’s 40 minutes and 32 kcal over the woman’s 30.
The second feature is the minus sign on weight in the women’s column. It looks like a typo and isn’t. The paper’s table of estimates gives −0.1263 with a standard error of 0.1061, so the study couldn’t tell that term from zero. For women, pulse told the model nearly everything and weight added little.
Its reference was total oxygen use run through Weir’s equation, and the paper doesn’t say resting energy was subtracted. It calls the output “physical activity energy expenditure”, and Chiu’s team added a separate resting estimate. Whether the answer includes your sitting-down share is unsettled. I treat it as a total. The daily version is in how many calories you burn a day.
How close did the formula come in its own test?
Close enough to compare one session with another, and not close enough to settle a food budget. In the 115 people used to build it, the model without fitness had a correlation of 0.857 with the measured burn. That means it accounted for 73.4% of the variation. Adding VO2 max lifted the figure to 83.3%. Both scores describe the people the equations were fitted to, so a new group should fit worse. It did.
The harder test used 17 new exercisers, 9 men and 8 women aged 21 to 53, who did 20 minutes of cardio they picked themselves. For the model without fitness the correlation fell to 0.77. On average the estimate ran 6.27 kJ a minute high, which is 1.5 kcal. The paper’s 95% limits of agreement ran from 12.65 kJ a minute too low to 25.19 too high. In kilocalories, that is 3.0 too low to 6.0 too high. That is wide.
Apply those limits to the three sessions. If the paper’s limits held for you, the real burn would sit between the estimate minus 6.0 kcal a minute and the estimate plus 3.0. That should hold about 95% of the time.
| Session | Formula estimate | Range from the paper’s limits | Width of range |
|---|---|---|---|
| Man, 38, run, 40 min | 607 kcal | 366–728 kcal | 362 kcal |
| Woman, 29, run, 30 min | 278 kcal | 97–369 kcal | 272 kcal |
| Woman, 44, brisk walk, 60 min | 415 kcal | 78–596 kcal | 518 kcal |
The ranges are my arithmetic on the paper’s limits, scaled from 20 minutes to longer sessions by my assumption. The last row needs a note. Its low end would be 53 kcal. But a 78 kg person uses about 78 kcal in an hour of sitting. The Compendium of Physical Activities sets 1 kcal per kg per hour as the resting rate, so I raised the floor to that. Seventeen people for 20 minutes is a small base, so read every range as rough.
All 115 volunteers exercised regularly. None had a known heart or metabolic condition or took medication every day. If that isn’t you, ask your doctor about exercise and pulse. A calculator can’t answer it.
When does a pulse overstate the burn?
When the pulse climbs and the work doesn’t. Long steady sessions, heat and dehydration are the usual causes.
In a study of nine male cyclists, riding at a constant effort in 35 °C, heart rate rose 12% between minute 15 and minute 45. That was 151 to 169 beats per minute. Oxygen use at that pace rose only slightly. Researchers call this cardiovascular drift. Their riders’ maximum oxygen uptake fell 19% over the same half hour. So the ride took a bigger share of what they could do, but its energy cost stayed nearly flat. A pulse-based formula would count those extra beats as extra burn (my reading, not the paper’s).
A 2003 review of heart rate monitoring agrees. Pulse climbs steadily during exercise in most of the studies it covers. Dehydration and air temperature have a profound effect on the link between pulse and oxygen use. On turning pulse into energy, its verdict is short: satisfactory for a group, not very accurate for one person.
Coffee gets blamed too. So I checked. A 2018 meta-analysis pooled 26 trials. Each gave 3–6 mg of caffeine per kg of body weight before steady exercise at 60–85% of maximum oxygen uptake. It found no effect on heart rate, and none on oxygen use. At those doses and efforts, coffee isn’t the likely culprit. Stress and nerves also raise pulse. I found no study I could open that puts a calorie figure on them, so there’s none here.
The formula can also fail when you’re outside its range. It was built on steady stages of 10 to 15 minutes at 57–90% of maximum heart rate. The authors say equations are still needed for intermittent activity and for lower intensities.
To see whether your pulse sits inside the range, estimate your maximum with 208 minus 0.7 times your age. That result comes from an analysis of 18,712 people. Multiply by 0.57 and by 0.90. For the three sessions above the ranges come to about 103–163, 107–169 and 101–159 beats per minute, and all three pulses fit. A slow stroll at 90 beats per minute doesn’t fit. Neither do weight sets with rests in between.
Before you use the formula, ask whether the session was steady and cool, and not far past an hour. If it was long or hot, or if you kept stopping and starting, read the answer as probably high.
How far off can a watch be?
Far off for calories, and not far for pulse. In 2017, Shcherbina and colleagues tested seven wrist devices on 60 adults. The adults sat, walked, ran and cycled, with an ECG and a breath-gas analyzer as the references. The Apple Watch had the lowest pulse error, a median of 2.0%. No device got its energy error under 20%. Median energy errors by device ran from 27.4% to 92.6%.
Newer evidence agrees. A living review from 2026 pooled 82 studies of the Apple Watch. For pulse, the pooled limits of agreement were −7.19 to +6.64 beats per minute. Energy was another story. All six studies that reported a mean absolute percentage error found 20% or more in at least one condition. The values ran from 9.71% for running to 151.66% for walking. Five of the eight energy studies used a Series 2 watch or older, so the evidence trails the hardware.
Look at what that does to the formula. A pulse error of 7 beats per minute is about 1.1 kcal a minute in the men’s equation (7 × 0.15). Over 40 minutes that is 42 kcal. It’s small beside the 9 kcal a minute between the formula’s own limits. The weak step is the pulse-to-energy conversion; the pulse reading itself is sound. Shcherbina’s team noted that the watch calculations are proprietary and suspected that some now use heart rate. So a watch’s calorie figure is a model you can’t inspect, sitting on a pulse reading you can trust more.
Device-by-device results are in how accurate Apple Watch calorie burn is. The same question for a whole day’s budget is in how accurate TDEE calculators are. Check your own pulse trace after a steady session instead: spikes and gaps mean any calorie figure sits on bad input.
How does the heart rate method compare with METs?
The MET method ignores your pulse and asks what you did. A MET is a multiple of resting energy use, and the Compendium sets 1 MET at 1 kcal per kg per hour. So calories equal the MET value times your weight in kilograms times hours. The 2024 Adult Compendium lists 1,114 activities for ages 19 to 59, and 912 carry values from measured oxygen use. Its editors left out studies that estimated energy from heart rate monitors. The two methods rest on separate evidence.
Here are the three sessions through both. The MET values are the Compendium’s for running and for brisk walking, with the activity code in brackets. I paired each pace with a pulse, and that pairing is my assumption.
| Session | MET value (code) | MET estimate | Heart rate estimate | Heart rate vs MET |
|---|---|---|---|---|
| Man, 38, 82 kg, run at 6–6.3 mph, 40 min | 9.3 (12050) | 508 kcal | 607 kcal | +19% |
| Woman, 29, 62 kg, run at 5.0–5.2 mph, 30 min | 8.5 (12030) | 264 kcal | 278 kcal | +5% |
| Woman, 44, 78 kg, walk at 3.5–3.9 mph, 60 min | 4.8 (17200) | 374 kcal | 415 kcal | +11% |
The heart rate method came out higher all three times. That fits the paper’s own check, where the model ran 1.5 kcal a minute high on average. Three invented sessions prove nothing, though.
Each method misses in its own way. A MET value is an average over many people, so it can’t tell if you’re fitter or less fit than most. The Compendium’s authors say as much: it doesn’t give precise individual energy values. A pulse does know something about you, but it also reacts to heat and drift.
When the two figures agree within about 10%, I’d trust both a little more. When they split by 20% or more, look for the reason, such as a hot day or a long session. A pace table is in how many calories running burns, and walking by distance and speed is in calories burned walking.
How do you check your own watch against the formula?
Compare three numbers from one steady session: the watch’s, the formula’s and the MET figure.
- Pick a steady run, ride or brisk walk of 30 to 60 minutes. Skip intervals and very hot days.
- From the watch’s workout summary, write down the average heart rate and the minutes. Add your weight in kilograms and your age.
- Work out the Keytel figure for your sex from the table above.
- Find the Compendium value for what you did, then multiply it by kilograms and hours. For a walk, the steps to calories calculator gives a figure from your step count and weight to set beside the others.
- Put the watch’s total next to the other two. If it shows only active calories, add the resting share first, as the total-energy reading above implies: your weight in kilograms times hours, times 1 kcal per kg per hour. For the man of 38 that is 55 kcal.
If all three sit within about 20% of each other, take the middle one as a working figure for that kind of session. That cut-off is mine, not a published one. If they scatter wider, no single number deserves much faith, and it stays out of your food plan.
Use the figure for one job: seeing whether this week’s session was harder than last week’s, with the same inputs. It is not permission to eat more. The range for the 40-minute run was 362 kcal wide, which is more than half the estimate. That is a poor base for a food decision.
Pulse is the one input here you can measure directly, and the Keytel formula is a fair way to turn it into a figure. Its error is still wide, so I’d keep it as a yardstick for comparing sessions. A week’s real cost shows up elsewhere: in your food log and, over a few weeks, in the trend of your weight.
Questions and answers
Is the heart rate formula closer to the truth than my watch?
The studies I opened never tested the two against each other, so I can't rank them. The formula's check on 17 people ran 3.0 kcal a minute too low to 6.0 too high. Seven wrist devices had median energy errors of 27.4% to 92.6% in a separate 60-person test. Different people, different protocols.
What heart rate range does the Keytel formula cover?
Its three steady stages sat at about 57%, 77% and 90% of maximum heart rate. One estimate of maximum, from a 2001 analysis of 18,712 people, is 208 minus 0.7 times age, so for a 38-year-old the range runs from 103 to 163 beats per minute. Below it, the formula is stretching past its data.
Does coffee push a heart rate calorie estimate too high?
Not in a 2018 meta-analysis of 26 trials. Caffeine at 3-6 mg per kg of body weight had no effect on heart rate or oxygen use during steady exercise at 60-85% of maximum oxygen uptake. Long sessions and heat are likelier reasons for a pulse that runs high.
Does the formula work for weight training?
It wasn't built for it. The study used steady running and cycling stages of 10 to 15 minutes, and its authors say equations are still needed for intermittent activity. Pulse jumps and falls between sets, which a formula fitted to steady stages was never tested on.
Sources
- Keytel LR, et al. Prediction of energy expenditure from heart rate monitoring during submaximal exercise. J Sports Sci 2005;23(3):289-297, Journal of Sports Sciences
- Jung WS, et al. Estimating excess post-exercise oxygen consumption using multiple linear regression in healthy Korean adults: a pilot study. Phys Act Nutr 2021 (reprints both Keytel equations), Physical Activity and Nutrition (PMC)
- Chiu YW, et al. Day-to-day variation of the heart rate, heart rate variability, and energy expenditure during FIFA 11+ and dynamic warm-up exercises. J Hum Kinet 2022 (reprints the Keytel equation without VO2 max), Journal of Human Kinetics (PMC)
- Unit Conversions (1 MET = 1 kcal/kg/hr), Compendium of Physical Activities, Compendium of Physical Activities
- Wingo JE, et al. Cardiovascular drift is related to reduced maximal oxygen uptake during heat stress. Med Sci Sports Exerc 2005, Medicine and Science in Sports and Exercise (PubMed)
- Achten J, Jeukendrup AE. Heart rate monitoring: applications and limitations. Sports Med 2003, Sports Medicine (PubMed)
- Glaister M, Gissane C. Caffeine and physiological responses to submaximal exercise: a meta-analysis. Int J Sports Physiol Perform 2018, International Journal of Sports Physiology and Performance (PubMed)
- Tanaka H, Monahan KD, Seals DR. Age-predicted maximal heart rate revisited. J Am Coll Cardiol 2001, Journal of the American College of Cardiology (PubMed)
- 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, Journal of Personalized Medicine (PMC)
- Lambe R, et al. The accuracy of Apple Watch measurements: a living systematic review and meta-analysis. npj Digit Med 2026, npj Digital Medicine (PMC)
- Herrmann SD, et al. 2024 Adult Compendium of Physical Activities: a third update of the energy costs of human activities. J Sport Health Sci 2024, Journal of Sport and Health Science (PMC)
- Running, code 12050 (9.3 METs) and 12030 (8.5 METs), 2024 Adult Compendium of Physical Activities, Compendium of Physical Activities
- Walking, code 17200 (4.8 METs), 2024 Adult Compendium of Physical Activities, Compendium of Physical Activities



