Why Your Fitness Tracker Overestimates Calories Burned
Your watch says you burned 620 calories. Stanford researchers who validated seven wrist-worn devices found that none of them estimated energy expenditure within an acceptable error range — even while measuring heart rate almost perfectly. Here is what is going wrong, and what to do about it.
In this guide
What the validation studies found
The most quoted work here is a 2017 study from Stanford’s Department of Medicine, which tested seven wrist-worn devices against laboratory measurement — indirect calorimetry for energy expenditure, ECG for heart rate — across walking, running and cycling. The headline result is worth reading slowly: six of the seven measured heart rate within an error of five percent, and none of the seven measured energy expenditure acceptably. Median error across devices ranged from roughly 27 percent to over 90 percent depending on the device and the activity.
That pattern has repeated across subsequent research. Trackers are decent at counting steps, good at heart rate at rest and steady state, moderate at sleep staging, and poor at calories. The one number most people actually make decisions with is the least reliable thing the device produces.
Put it in context
A 30 percent error on a reported 600-calorie workout is 180 calories — about a third of a typical daily deficit. A 90 percent error is more than the entire deficit. Eating back an overestimated burn is one of the most common reasons a diet quietly stops working.
Why energy expenditure is so hard to estimate
It is not laziness on the manufacturers’ part. The problem is that your wrist cannot see what actually determines your energy cost.
- Heart rate is not a calorie meter. The relationship between heart rate and oxygen consumption is individual, and it shifts with caffeine, heat, dehydration, sleep debt, stress and training status. Two people at 150 bpm can be burning very different amounts, and the same person at 150 bpm can differ from one day to the next.
- Resistance training breaks the model. A heavy set spikes your heart rate through pressure response and neural drive, not sustained aerobic work. Trackers read the spike as cardio and bill you accordingly. Lifting is one of the activities where overestimation is worst.
- Body composition is invisible to it. Your device knows weight, age, sex and height. It does not know your lean mass, and lean mass is what drives resting expenditure — which is why two people of identical weight and very different composition get the same estimate.
- Proprietary algorithms are unvalidated. Each manufacturer applies its own undisclosed model. You cannot check the arithmetic, and it can change silently in a firmware update.
- Wrist optics have limits. Photoplethysmography reads blood flow through skin, and it degrades with motion, skin tone, tattoos, fit, and cold hands. Errors in the input compound in the output.
The double-counting problem
Even a perfectly accurate device would mislead you, because of an accounting error almost every app makes.
The calories your body spends over an hour include the calories you would have spent anyway just being alive. An hour of moderate cycling at 8 METs costs about 8 times resting metabolism — but roughly one of those MET-hours is your baseline, which is already inside the TDEE figure your diet is built on. When your tracker adds the full gross burn to your daily budget, it is counting that baseline twice.
| Figure | 80 kg person, 1 hour at 8 METs | What it means |
|---|---|---|
| Gross expenditure | ~640 kcal | Total energy spent during the hour |
| Resting baseline | ~80 kcal | Already included in your TDEE |
| Net exercise burn | ~560 kcal | The calories the workout actually added |
| Typical app credit | 640 kcal or more | Double-counts the baseline, then adds algorithm error on top |
Subtract one MET from any activity value before crediting yourself for it. Our MET values table lists the gross figure for 180-plus activities, so the correction is easy: burn per hour is roughly MET minus one, times your bodyweight in kilograms.
What your tracker is genuinely good at
None of this makes the device worthless. It makes it a behaviour tool rather than an accounting tool, and it is very good at that job.
| Metric | Reliability | Use it for |
|---|---|---|
| Step count | Good | Holding a daily movement floor — the best defence against NEAT collapse in a diet |
| Resting heart rate | Good | Trend tracking; a rising baseline often flags fatigue, illness or poor sleep |
| Heart rate, steady state | Good | Pacing training zones |
| Heart rate, intervals | Moderate | Rough zone feedback; lags fast changes |
| Sleep duration | Moderate | Week-to-week trends, not single nights |
| Calories burned | Poor | Comparing your own workouts to each other — never for eating back |
The crucial distinction is relative versus absolute. If your watch says today’s run burned 20 percent more than last week’s, that comparison is probably meaningful, because the error is reasonably consistent for the same person doing the same activity. The absolute number is what you cannot trust.
How to use a tracker without wrecking your diet
- Do not eat back exercise calories. The simplest fix. Set your intake target from a TDEE calculation that already includes an activity multiplier, and treat exercise burn as a bonus you never spend.
- If you must eat some back, take half. Halving the credit lands you closer to the net figure once double-counting and algorithm error are accounted for.
- Use steps as your activity metric, not calories. A step floor of 8,000 to 10,000 a day is measurable, unambiguous, and directly targets the spontaneous movement that quietly disappears during a deficit.
- Let the scale be the arbiter. Weigh daily, average weekly, and compare three-week trends. Two to three weeks of your own data beats every estimate a device or a formula can produce.
- Recalculate your target when your weight changes by 5 kg. A lighter body burns less, and your tracker’s baseline assumptions drift with it.
One reframe that helps. Your tracker is not measuring your metabolism — it is estimating it from a wrist. Treat its calorie number the way you would treat a stranger’s guess about your salary: probably in the right order of magnitude, definitely not something to budget against. The scale, a tape measure, and three weeks of patience are your real instruments. Everything else is a nudge to move more, which is what these devices are genuinely excellent at.
Common questions
The honest answer is that none of them are accurate enough to eat back. The 2017 Stanford validation found no device among seven measuring energy expenditure within an acceptable error range, despite most reading heart rate well. Chest-strap heart rate improves the input but does not fix the underlying modelling problem.
Two reasons usually stack. Wrist algorithms tend to run high, especially for resistance training where heart rate spikes without corresponding aerobic work. And most apps report gross expenditure, which includes the resting calories already counted inside your TDEE.
As a default, no. Set your intake from a TDEE figure that already includes an activity multiplier for your typical week. If you train far more than usual in a given week and feel genuinely under-fuelled, eating back about half the reported burn is a reasonable compromise.
For heart rate, yes — chest straps measure electrical activity directly and are considerably more accurate during intervals and high-intensity work. For calories, they improve the input to a model that still cannot see your lean mass or metabolic individuality, so the estimate remains an estimate.
Generally yes, within roughly 5 to 10 percent for normal walking, which is easily good enough for its purpose. Steps are the metric most worth acting on, because holding a daily floor prevents the drop in spontaneous activity that derails most diets.
Last reviewed September 14, 2026. Sources are cited inline; see our editorial policy for how we choose and review them.