No consumer wrist tracker has yet shown that it can measure calorie burn with real precision. The most-cited test, a Stanford-led 2017 study of seven popular devices, found that heart rate readings were mostly decent while energy expenditure estimates were not. The best performer was out by around 27% on average, and the worst by more than 90%.
Those are not rounding errors.
- No fitness tracker achieves better than 27% accuracy for calorie expenditure; most miss by 50% or more.
- Heart rate readings are reasonably reliable across brands, but energy expenditure estimates are the weakest measure by far.
- Cycling, weight training and interval workouts trip up calorie algorithms in different ways depending on wrist movement and heart rate patterns.
What the lab studies found
The study, published in the Journal of Personalized Medicine, put 60 volunteers in wrist-worn devices including the Apple Watch, Fitbit Surge and Samsung Gear S2. Participants walked, ran and cycled while researchers compared the wearables against lab-grade measures: ECG for heart rate and indirect calorimetry (analysing the gases you breathe out) for energy use. Most devices managed heart rate within about 5%. None came within 20% for calories.
Those hardware generations are now several years old, and you could reasonably argue that a 2026 watch should do better. A 2020 systematic review in JMIR mHealth and uHealth, which pooled many validation studies across brands, points the same way. Heart rate was fairly dependable, step counts tended to run low, and energy expenditure was the weakest measure by a distance.
A 25% error on a 400-calorie workout is 100 calories, roughly a medium banana. Over a week of sessions that gap stops being trivial.
Brand by brand: a messier picture than the marketing
Apple and Fitbit have the deepest research base, simply because so many people wear them and so many university teams have tested them. Garmin, Polar and Samsung appear regularly too. What stands out is how often the league table reshuffles: a device that tops one study lands mid-pack in the next, depending on the model tested, the firmware version and the activities chosen.
That inconsistency is itself the finding. If one brand were reliably better at calories, the studies would keep crowning it, and they do not.
Newer categories, such as smart rings and screenless straps, have thinner independent evidence on calorie accuracy. Their makers often publish internal validation, but I would treat that as a starting claim, not a verdict. A device with fewer peer-reviewed tests is not necessarily worse, only less proven.
Why the number wobbles
A tracker does not measure calories. It estimates them from heart rate, wrist movement and the profile you entered (age, sex, height, weight), then feeds those into an equation built on averages. If your fitness, body composition or heart rate response differs from the average, the equation drifts.
Activity type matters more than the logo on the strap. Cycling barely moves your wrist, so motion-based estimates suffer. Weight training raises energy use without sustained heart rate spikes, which leaves the model guessing. Intervals, stairs and carrying loads all trip up the algorithm in different ways, and the direction of the error (over or under) changes by device and by activity.
The all-day total is shakier still. Most of what you burn is resting metabolism, which the device also estimates from an equation, so a “2,600 calories today” figure carries uncertainty from top to bottom.
Using the number sensibly
Treat the figure as a trend line, not a ledger. A tracker that is consistently 20% high is still useful for seeing whether this week was busier than last, because the bias cancels out when you compare like with like. Keep your profile details current, wear the device snugly, and consider a chest strap for heart rate if you do a lot of intense training, as these are generally more reliable than wrist optical sensors.
Eating back every calorie your watch reports is the habit most likely to backfire. If weight loss or maintenance matters to you, check the maths against reality: log your average weight over three to four weeks and see whether it matches what the numbers predicted.
People with a history of disordered eating, or with a medical condition that affects diet or exercise, should talk to their GP before using calorie data to steer decisions. Numbers that look precise can feel authoritative when they are not.
One last wrinkle: the other half of the equation is shaky too. Research using doubly labelled water, a gold-standard method for measuring real energy expenditure, has repeatedly shown that people under-report what they eat in food diaries, so a tracker error and a logging error can stack in either direction.