Short answer: your smartwatch measures heart rate well and calorie burn poorly. In Stanford's widely cited accuracy study, six of seven wrist wearables tracked heart rate within 5% error, but even the *best* device was off on energy expenditure by a median of 27%, and the worst by 93%. So if you suspect your Apple Watch or Fitbit burn number is wrong, you're probably right. That doesn't make the watch useless. It means you should treat the number as a rough, consistent signal, not a fact you can eat against.
What the research actually found
The benchmark study is Shcherbina et al. (2017) from Stanford, published in the *Journal of Personalized Medicine*. Researchers put seven wrist devices, including the Apple Watch, Fitbit Surge, Samsung Gear S2, and Microsoft Band, on 60 volunteers while they sat, walked, ran, and cycled. They compared the devices against clinical gold standards: ECG for heart rate and indirect calorimetry (measuring oxygen and carbon dioxide in your breath) for calorie burn.
The results split cleanly in two:
- Heart rate: six of seven devices stayed under 5% median error. Genuinely good.
- Energy expenditure: the most accurate device (Fitbit Surge) had a median error of 27.4%. The least accurate (PulseOn) was off by 92.6%. No device came close to the 10% error the researchers considered acceptable.
This isn't one outlier study. A 2020 systematic review in *JMIR mHealth* (Fuller et al.) pooled 158 publications covering Apple, Fitbit, Garmin, Samsung, Polar, and others. Its conclusion on calorie burn was blunt: "For energy expenditure, no brand was accurate." Steps and heart rate held up reasonably well; calories did not.
Devices have improved since 2017, but the physics of the problem hasn't changed, which is why newer validation studies keep finding the same pattern.
Why the estimates are off
Your watch never measures calories. It measures motion and heart rate, then runs them through a prediction model. Every step in that chain adds error.
Your baseline metabolism is a guess
The watch estimates your basal metabolic rate from age, sex, height, and weight. Real BMR varies meaningfully between people with identical stats, differences in muscle mass, hormones, and genetics can shift it by a couple hundred calories per day. If the baseline is wrong, everything built on it is wrong.
Heart rate is not oxygen consumption
Calorie burn tracks oxygen consumption, and heart rate is only a proxy for that. Stress, caffeine, heat, dehydration, poor sleep, and fitness level all push your heart rate up or down without changing how much energy you're actually burning. A nervous meeting can read like a brisk walk.
The watch has to guess what you're doing
Wrist sensors detect arm movement. Cycling barely moves your wrist. Carrying groceries looks like nothing. Pushing a stroller pins your arm in place. When the device misclassifies the activity, it applies the wrong burn model entirely.
Algorithms are tuned to averages
Manufacturers train their models on test populations. If your body composition, efficiency, or movement style differs from that average, and everyone's does in some direction, the error is baked in and systematic. That's why your watch tends to be consistently high or consistently low rather than randomly scattered.
The worst activities (and the best)
Error isn't evenly distributed. In the Stanford data and follow-up research, accuracy depends heavily on what you're doing:
Least accurate:
- Cycling — minimal wrist motion, so the watch leans almost entirely on heart rate.
- Strength training — short efforts spike your heart rate out of proportion to oxygen use, and rest between sets confuses the model.
- HIIT — heart rate lags behind effort during rapid intervals, so the watch is always estimating from stale data.
Most accurate:
- Steady-state running and walking — rhythmic, wrist-visible, heart rate stabilized. This is what the algorithms were trained on, and it shows.
If most of your exercise is lifting or intervals, assume your burn number is on the flattering side of wrong.
How to actually use the number
The watch's real strength is *consistency*, not accuracy. It measures you the same wrong way every day, which makes it a solid trend tool:
- Compare you to you. If your daily burn is usually 550 and this week it's 400, you genuinely moved less. That comparison is valid even if both numbers are inflated.
- Use it for motivation and habit. Closing rings works because it's a consistent behavioral target, not because 630 active calories means 630 calories.
- Never transfer the raw number into your food budget. Eating back 500 watch-calories that were really 320 quietly erases your deficit, this is one of the most common reasons the scale won't move. We break down the math in should you eat back exercise calories.
- Know which number you're looking at. Active and total calories answer different questions, and mixing them up doubles the damage. Here's active calories vs total calories explained.
How to calibrate your watch against reality
You can find out exactly how wrong your watch is with three weeks of honest data:
- Log your food intake accurately for 2-3 weeks, weighing portions where you can.
- Weigh yourself daily and use the weekly average, day-to-day water swings are noise.
- Do the math. Each kilogram of body weight change represents roughly 7,700 kcal. If you ate 2,300/day, your watch claimed you burned 2,800/day, but your weight held steady, your true burn is about 2,300 and the watch runs ~500 high.
- Adjust your targets to the real number, not the watch's. Our TDEE calculator gives you a research-based starting estimate to compare against.
Your logged intake plus your weight trend *is* a calorimeter. It's slower than a watch, but it's the one measurement that can't lie about energy balance.
The bottom line
Your Apple Watch, Fitbit, or Garmin is a good heart rate monitor, a good step counter, and a mediocre calorie estimator. Trust the trends, ignore the absolutes, and let your food log and scale settle any argument between them.
Healthy Buddy makes that calibration loop easy: log meals in seconds, watch your weight trend, and see your *real* energy balance instead of your watch's guess. Download Healthy Buddy free.
FAQ
Should I lower my Apple Watch calorie goal?
Not necessarily. The goal works as a consistency target even if the number is inflated. Only adjust it if you're using the burn figure to justify extra food, in that case, fix the food budget instead of the ring.
Is Fitbit or Apple Watch more accurate for calories?
Neither is reliably accurate. In the Stanford study the Fitbit Surge had the lowest energy-expenditure error (27.4% median) among the seven devices tested, while the Apple Watch had the most accurate heart rate. Both are fine for trends; neither is fine for food math.
Why does my watch say I burned 800 calories in a 45-minute workout?
Probably a mix of inflated exercise estimates and total calories (which include your baseline metabolism for that time window) being read as extra burn. For strength training or HIIT specifically, heart-rate-based estimates run high. Halve it mentally before it goes anywhere near your food log.