Smartwatch Calorie Accuracy Drops as Body Fat Rises, Study Finds

Millions of people glance down at their wrist after a workout and trust the number staring back at them: 350 calories burned, 500 calories burned, whatever the little screen says. That number often decides whether someone eats an extra snack, skips dessert, or feels good about a hard workout. A new study suggests that number might be wrong, and the error grows with how much body fat a person carries.

Researchers at Florida International University put four popular smartwatches through a controlled cycling test and compared their calorie estimates to a medical-grade breath test that measures energy burned with real precision. Every device missed the mark to some degree. But the more important finding, published in the journal PLOS One, went beyond simple inaccuracy: the error grew with a person’s body fat percentage, no matter which brand was strapped to the wrist.

That detail matters because people managing their weight, often carrying more body fat, are exactly the population leaning hardest on these devices to plan meals, track exercise, and make health decisions. If the tool is least accurate for the people relying on it most, that’s a problem worth taking seriously.

How Researchers Tested Smartwatch Calorie Accuracy

A group of 58 Hispanic adults between ages 18 and 50 took part, deliberately chosen to include people across a range of body sizes and skin tones, from medium to deep. Nearly half the group had a body mass index of 30 or higher, meaning they were classified as obese, allowing the researchers to test whether body composition changed the results.

Each participant pedaled a recumbent exercise bike for about ten minutes, alternating between two-minute bursts of moderate and harder effort, with five-minute rest periods on either end. During the session, everyone wore four smartwatches at once: an Apple Watch Series 8, a Fitbit Sense 2, a Samsung Galaxy Watch 5, and a Garmin Forerunner 955. At the same time, they were hooked up to a portable machine that measures the oxygen and carbon dioxide in a person’s breath to calculate exactly how many calories they burned. That breath-based reading served as the benchmark against which each smartwatch’s estimate was measured.

Researchers also measured each participant’s body fat percentage with a handheld device and recorded their skin tone using a standard scale doctors use to classify skin types.

Where Each Smartwatch Went Wrong on Calories Burned

Three of the four watches showed errors that leaned consistently in one direction rather than scattering randomly. Garmin overestimated calories burned by about 69 kcal per session on average, and Samsung overestimated by about 57 kcal. Apple’s overestimate was smaller, averaging about 22 kcal.

Fitbit’s results looked good at first glance, with an average bias of just over 3 kcal. But that number hid a messier problem. Fitbit produced seven readings so extreme, more than four and a half times the true calorie value, that researchers had to pull them out as errors before running the main analysis. Those outliers showed up in about 13% of Fitbit’s trials. When those extreme numbers were left in, Fitbit’s average bias jumped to nearly 129 kcal, by far the largest miss of any device tested. Fitbit also failed to produce a reading at all in one session.

Beyond the direction of the errors, researchers also looked at how far off each device tended to be overall. Apple and Fitbit generally landed closer to the true number, while Garmin and Samsung tended to be off by more.

Body Fat Skews Smartwatch Calorie Estimates, Not Skin Tone

Accounting for body fat percentage changed the picture. As body fat rose, calorie estimate errors rose with it, and this held true across all four brands. The size of that increase varied by device and by how researchers measured the error, though no brand was exempt from the overall body-fat pattern.

Skin tone told a different story. Despite earlier research suggesting that darker skin might interfere with the light-based sensors these watches use to detect heart rate, this study found no clear connection between skin tone and calorie-estimate accuracy. The researchers cautioned that this result should be read carefully, since very few participants in the study had the darkest skin tone category examined, which limited how confidently that conclusion could be drawn.

Those same light-based sensors work by shining light into the skin and tracking how blood flow changes what bounces back. Earlier research cited in the study found that body composition can interfere with that signal even more than skin pigmentation does, and this new data lines up with that idea.

Why This Matters Beyond the Gym

Smartwatch calorie counts feed into many real decisions. People use them to figure out how much they can eat during a diet, doctors sometimes reference activity data in clinical conversations, and researchers use wearable data to study exercise and health trends across large groups. If the accuracy of that data gets worse for people with higher body fat, a group already at greater risk for heart and metabolic problems, then the tool meant to help them could be sending them in the wrong direction without anyone realizing it.

Researchers behind the study were blunt in their conclusion: current consumer smartwatches aren’t reliable enough yet for tracking calories burned, whether for someone trying to manage their weight or for scientists studying large groups of people. One possible fix, the researchers say, is for companies to train their calorie-estimating formulas using data that better represent people across different body types, so the accuracy of these devices doesn’t quietly depend on who’s wearing them.

Disclaimer: This article summarizes findings from a peer-reviewed study and is provided for general informational purposes only. It is not medical or health advice. Do not use smartwatch calorie readings to make dietary, weight-management, or medical decisions without guidance from a qualified healthcare professional. Individual results may vary, and the study was conducted under controlled cycling conditions with a specific group of participants, so findings may not apply to everyone or to every type of activity.

---