Fitness Trackers Overestimate Calorie Burn by Up to 93 Percent. Here Is Why the Formula Has Always Had That Problem.
Consider someone training for a half marathon. She runs four mornings a week, her Apple Watch tells her she burned 520 calories, and she eats a larger lunch to recover. She is being careful. She is tracking everything. After three months, she has gained two pounds. The watch was not broken. The underlying formula has had a structural flaw for more than sixty years, and the device was faithfully reproducing it.
The problem is not a recent discovery. In 2017, a team at Stanford University tested seven popular fitness wearables on 60 volunteers. The researchers used medical-grade breath analysis to measure actual oxygen consumption, the gold standard for calculating energy expenditure. Every single device they tested was wrong. The most accurate was still off by an average of 27 percent. The least accurate was off by 93 percent. Heart rate, by contrast, was measured well: six of the seven devices came within 5 percent of the clinical electrocardiogram readings. The hardware was doing its job. The calorie estimates were not.
This article traces the problem to its source. The formula that underlies almost every calorie-burn calculation was calibrated decades ago on a very specific kind of person, and research has since shown that the calibration does not generalize. Understanding where the number comes from explains not just why trackers fail, but why the failure almost always runs in the same direction: too high.
The 40-Year-Old Man Behind Every Calorie Estimate
The unit that most calorie-burn formulas depend on is the MET, or Metabolic Equivalent of Task. One MET is defined as the amount of oxygen your body consumes at rest: 3.5 milliliters of oxygen per kilogram of body weight per minute. From there, activities are assigned multipliers. Walking at 3.5 miles per hour carries a MET value of roughly 3.5, meaning it costs 3.5 times more energy than sitting still. Running at 6 miles per hour is approximately 9.8 METs. Multiply the MET value by body weight in kilograms and duration in hours, and you arrive at an estimated calorie burn.
The framework is logical. The problem is the anchor.
That 3.5 ml/kg/min baseline was derived from measurements taken on a single subject: a 70-kilogram, 40-year-old man. The exercise physiologist Bruno Balke proposed this value in 1960 as a standardized resting reference point. The idea was to give researchers a simple, consistent denominator for comparing the energy cost of different activities across studies. Convenience was the goal. Universality was assumed.
The term metabolic equivalent and the activity-level scale that most fitness apps still use were not formalized until much later. Jetté and colleagues codified the MET classification system in a 1990 review, establishing the intensity levels that showed up in clinical guidelines and eventually in the algorithms behind consumer wearables. By then, the 3.5 baseline had been embedded in exercise science literature for three decades without serious challenge.
The challenge came from population studies. Research published in the Journal of Applied Physiology found that for a large and diverse sample of people, the 3.5 ml/kg/min resting value overestimates actual resting oxygen consumption by an average of 35 percent. Older adults tend to have lower resting metabolic rates. Women typically have lower rates than men of the same body weight because of differences in muscle-to-fat ratios. Sedentary individuals often have lower rates than trained athletes. The 70-kilogram 40-year-old man was not a neutral reference point. He was one data point promoted to a universal constant, and every calorie estimate downstream inherits his particular physiology.
How Devices Compound an Already-Flawed Foundation
A fitness tracker does not measure calorie burn directly. It measures heart rate and movement through an accelerometer. From those signals, it uses a proprietary algorithm to estimate how much oxygen you are consuming, and from that it infers calories. Every step in that chain introduces additional error on top of the flawed MET baseline.
The Stanford study illustrated this layering clearly. Lead researcher Anna Shcherbina noted that training a reliable calorie algorithm is fundamentally difficult because energy expenditure varies based on fitness level, height, weight, and the specific nature of the activity. The seven devices tested included the Apple Watch, Fitbit Surge, Microsoft Band, Mio Alpha 2, Basis Peak, PulseOn, and Samsung Gear S2. All of them used different underlying models. None of them landed close enough for clinical use.
Activity type matters more than most users realize. Walking and running algorithms are better calibrated because these movements are well-studied and the arm-swing patterns are distinct. Cycling proved far harder. In the Stanford data, calorie errors for stationary cycling averaged around 52 percent, compared to roughly 31 percent for walking and running. The seated posture during cycling disrupts the arm-motion signals and alters the relationship between heart rate and energy expenditure in ways the algorithms do not account for well.
Senior author Euan Ashley summarized the results bluntly: "The heart rate measurements performed far better than we expected, but the energy expenditure measures were way off the mark." The finding was not that fitness trackers are useless. They are genuinely reliable for the thing they measure most directly, which is heart rate. They are unreliable for calorie estimation, and that unreliability comes from combining an imprecise six-decade-old baseline with the practical limits of inferring oxygen consumption from a wrist sensor.
The Compensation Effect That Makes Overstated Numbers Worse
The arithmetic problem with overestimated calorie burns becomes a behavioral problem when those estimates shape eating decisions. Research on exercise-related compensation has documented a consistent pattern: people unconsciously increase food intake on days when they exercise, independent of any deliberate attempt to eat back calories. Studies have found that active individuals burn roughly 28 percent fewer net calories than their tracker predicts because of this compensatory behavior, with the body partially offsetting exercise output through subtle reductions in resting metabolism and incidental movement throughout the rest of the day.
For individuals at higher body weights, the compensation effect is more pronounced. Research suggests these individuals recoup close to half of their actual calorie expenditure through reduced baseline activity after exercise.
Layer device overestimation on top of natural compensation and the effect compounds. Take someone who burns 400 calories on a run. Her tracker reports 560. She logs the 560 and eats back most of it. She has eaten back more than she actually burned before compensation is even factored in. On three workout days per week, this gap can accumulate to 500 or more extra calories weekly, which over several months produces real weight gain for someone who believed she was running a deficit the entire time.
The practical correction that exercise scientists most often suggest is to treat tracker calorie figures as directional rather than precise. If you plan to eat back exercise calories at all, eating back roughly half of the stated estimate is a more conservative and better-calibrated approach. The second half of the number is where the accumulated error lives.
Conclusion
None of this means calorie tracking is worthless. Consistency still has value. If your tracker reports 400 calories on a Monday run and 430 calories on the same route two weeks later, that comparison is meaningful. You put in slightly more effort, or you worked at a higher intensity. The trend is real even when the absolute total is off.
The more important lesson is to understand what you are looking at. A calorie burn total from a consumer fitness tracker is an estimate built on a convention established in 1960 from a single measurement, layered with device-specific algorithms that perform reasonably for heart rate and poorly for energy expenditure. It is a starting point for a decision, not a fact to log without skepticism.
If you want to work with the underlying MET math directly, a calorie burn calculator lets you input your own weight, activity MET value, and duration without the additional layers of device error. The formula still carries its original limitations, but using it consciously, with realistic expectations about what resting metabolic rate means for your specific body, puts you in a better position than trusting a number generated by a wrist device whose calorie algorithm has never been independently validated to within 20 percent.
Frequently Asked Questions
What is a MET value in exercise science?
MET stands for Metabolic Equivalent of Task. One MET is the energy your body uses at rest. Activities are rated by multiples of that baseline, so a 5 MET activity burns five times your resting rate.
Why do fitness trackers overestimate calories burned?
The MET baseline was set using a 40-year-old 70kg man. Real resting metabolic rates vary by age, sex, and fitness level, and devices compound the error with population-level heart rate models.
Should you eat back the calories your tracker says you burned?
With caution. Tracker estimates often run high, and research on the compensation effect shows exercisers frequently out-eat their actual calorie burn when relying on device readings.