Your Apple Watch says you burned 612 calories on that run. Your Garmin says 547. Your Whoop says 489. They can't all be right, and the research suggests none of them are. This guide explains why wearables miss the mark, by how much, and which numbers to actually trust when setting a calorie target.
The short version
Stanford tested seven leading wrist wearables against gold-standard indirect calorimetry. Heart rate was reasonably accurate (within 5%). Energy expenditure was not, the best device missed by 27%, the worst by 93%. Every single one tended to overestimate.
Why wearables overestimate
- They guess at your basal rate. Most use sex, age, height and weight plugged into the same Mifflin-St Jeor equation as the BMR calculator, then add an "active" multiplier on top. Anything beyond that is inference.
- Heart rate is a noisy proxy for calories. HR rises from caffeine, heat, dehydration, stress and poor sleep, none of which burn meaningful extra calories. The watch reads the spike as work.
- Strength training is the worst case. Lifting weights barely raises HR but does real metabolic work. Watches consistently undercount lifting and overcount steady-state cardio.
- "Active" calories vs "total" calories confuse people. Active is on top of BMR. Total includes it. Logging total in MyFitnessPal as an exercise burn double-counts BMR for the hours you were moving.
Which activities are least wrong
Steady-state cardio in a controlled environment, treadmill running, indoor cycling with a power meter, rowing on an erg, is where wearables do best, typically within 10-15%. Outdoor running with GPS adds another 5-10% of error. Lifting, HIIT, climbing, hiking with elevation, anything intermittent: error often exceeds 30%.
What to actually do with the numbers
- Don't eat back exercise calories. Build your daily target from TDEE using an honest activity multiplier and stop there. Eating back what the watch reported is the single most common reason "calorie deficit" diets stall.
- Use the watch as a relative meter, not absolute. If a typical run is 500 and today's says 700, that's useful directional info. The absolute number isn't.
- For exercise estimates, cross-check with a MET-based calculator. The calories burned calculator uses the same MET tables exercise physiologists use, and tends to undershoot watches by 15-25%, which is closer to truth.
- For heart-rate based training, trust HR, not calories. The heart rate zone calculator gives you targets that actually drive adaptation. Calories are a side effect.
What about the new ones, Whoop, Oura, Ultrahuman?
Newer recovery-focused wearables put more weight on HRV and resting HR than on activity. Their daily "burn" estimates are usually saner than older fitness watches, but still off by 15-20% in independent testing. The recovery score is more useful than the calorie number.
A working approach
Set calories from TDEE at the activity tier that matches your real week. Hit protein from the protein calculator. Ignore the watch's "ate vs burned" ring. Weigh in daily, average across seven days, and adjust calories based on what the scale actually does over 3-4 weeks. The bathroom scale is a more accurate calorimeter than any watch on your wrist.