Estimated Energy Expenditure and Calorie Burn
How watches and trackers estimate calories burned, what reference methods they are tested against, and why their calorie numbers are among the least accurate wearable outputs.
Content last updated . This page has not been medically reviewed.
Normal variability
Error is not constant. In the 2017 laboratory study, devices had the lowest error in cycling and the highest in walking, and error was higher for some groups of people. [1][1]What the numbers can and cannot support
- Appropriate uses: a rough sense that one day was more active than another. For movement itself, see step counts and perceived exertion.
- Not supported: deciding how much to eat from a watch's calorie number, or comparing "calories burned" between devices or people. With errors of 20% or more common in the lab, the number can mislead. [1]
- Cannot diagnose: a slow or fast metabolism, thyroid problems or the cause of weight change. See metabolism myths and resting metabolic rate estimates.
- Accuracy is not benefit. We did not read a source showing that tracking calories burned improves health or weight outcomes in this group of pages.
Who should be careful
People with a history of an eating disorder, or for whom calorie numbers cause anxiety or compulsive exercise, may wish to talk with a clinician before tracking. People with diabetes who use glucose-lowering medicine should not change medicine or food based on a watch estimate; talk with the prescriber.When professional assessment may be appropriate
Unintended weight change, new fatigue, a racing heart or breathlessness, rather than the calorie figure, are the reasons to see a clinician; see unintentional weight loss. For food-side numbers, see calorie estimates and label accuracy.Frequently asked questions
How accurate is the calorie count on my watch?
Poor in the studies we read. One lab study found no device with median energy-expenditure error under 20%, and reviews found no brand accurate for energy expenditure. [1][4]
Why is heart rate accurate but calories are not?
Heart rate is read from the body by a sensor, while calories are an estimate built on top. In the laboratory study, most devices measured heart rate adequately but estimated energy expenditure poorly. [1]
What is the best way to measure energy expenditure?
Research uses indirect calorimetry, metabolic chambers or doubly labeled water, which are not consumer methods. [2]
Should I eat back the calories my watch says I burned?
We found no evidence supporting that, and the errors above argue for caution. Ask a clinician or dietitian for individual advice.
More details: evidence, limits and cautions
The details behind this page: cautions, limits of the evidence and where sources disagree. Open any note to read it.
How to read the evidence labels
Labels used on this page: established, mixed, insufficient evidence, marketing claim.
What is measured and what is calculated
- Directly measured by the device: signals such as motion and heart rate. In one laboratory test, seven wrist devices were evaluated for estimating both heart rate and energy expenditure. [1]
- Calculated: energy expenditure, in kilocalories (kcal) per minute or per day. The studies we read test the outputs; the makers' formulas were not described in them.
- Reference methods: indirect calorimetry (breathing gases) in the lab, a metabolic chamber for a day, or doubly labeled water, which one study used over 15 free-living days. [1][2]
- Which "calories"? Total daily expenditure is made up of basal metabolic rate, the thermic effect of food and activity; non-exercise activity is everything we do that is not sleeping, eating or sports-like exercise. [3] The 12-device study focused on the activity part. [2]
Accuracy evidence
| Study | Design | Energy expenditure finding |
|---|---|---|
| Laboratory study (2017) | 60 volunteers, seven wrist devices of that time, indirect calorimetry; sitting, walking, running, cycling [1] | No device achieved a median error in EE below 20%; error higher with greater BMI, darker skin tone, in males and in walking [1][1] |
| Metabolic chamber and doubly labeled water | 19 healthy adults, 12 devices [2] | In free living, activity energy expenditure was significantly underestimated by all but two devices [2] |
| Systematic review (158 publications) | Commercial wearables | "For energy expenditure, no brand was accurate" [4] |
| Systematic review of one brand (67 studies) | Controlled and free-living | Unlikely to provide accurate energy expenditure in any testing condition [5] |
| Systematic review of wrist devices (65 articles) | General population | Mean absolute percentage error over 30% for all brands [6] |
| Meta-analysis, one smartwatch family (56 studies) | Bland-Altman limits of agreement | All EE subgroups exceeded the 10% validity threshold [7] |
These results describe specific devices and generations. Newer models may differ, and no model name is endorsed. A wide gap between a device and its reference is a measurement limit, not a flaw in the person using it.
References
- Shcherbina et al., J Pers Med 2017. Read 2026-10-22.
- Accuracy of 12 wearable devices for physical activity energy expenditure, JMIR Mhealth Uhealth 2019. Read 2026-10-22.
- Levine, Non-exercise activity thermogenesis (NEAT), Nutr Rev 2004. Read 2026-10-22.
- Fuller et al., JMIR Mhealth Uhealth 2020. Read 2026-10-22.
- Accuracy of Fitbit devices, systematic review, JMIR Mhealth Uhealth 2018. Read 2026-10-22.
- Wrist-wearable activity trackers, systematic review, J Med Internet Res 2022. Read 2026-10-22.
- Meta-analysis of Apple Watch accuracy, Physiol Meas 2025. Read 2026-10-22.