Quick Answer: Wrist-worn trackers estimate energy expenditure with a typical error of 20-30%, and studies have found individual errors above 90%. They infer calories from heart rate and motion using population averages that may not describe you. Step counts and heart rate trends are reasonably accurate; the calorie number is the least reliable thing on the screen.

Which does not make the device useless. It makes the calorie figure the wrong metric to build decisions on.

Use it as a consistency tracker, not an accounting ledger.

How wrong are tracker calorie estimates, really?

The research has been fairly consistent across device generations.

A well-known Stanford evaluation of seven consumer wrist devices found heart rate measurement was quite good - typically within 5% error - while energy expenditure estimates missed badly, with the best device off by about 27% and the worst by more than twice that.

Subsequent studies have found the same pattern: the sensor is fine, the model on top of it is guessing.

Why the split? Heart rate is directly measured through optical sensing. Calorie burn is calculated from heart rate, motion, and your entered height, weight, age, and sex, run through equations derived from population averages. If your resting metabolic rate, fitness level, or mechanical efficiency differs from that average - and most individuals differ somewhere - the output drifts.

The number is precise-looking and imprecise. 612 feels authoritative in a way that ‘roughly 450 to 750’ does not.

Which activities does it get most and least wrong?

Error is not uniform, which is useful to know.

Most accurate: steady-state cardio with consistent motion - walking, jogging, cycling at a stable pace. The motion signal and the heart rate signal agree, and the underlying model was largely built on this.

Moderately off: interval training, hiking with elevation, rowing. Heart rate lags actual effort by 30-60 seconds, so short intervals get systematically underestimated.

Badly off: strength training. Wrist trackers overestimate resistance training substantially - the arm is often nearly still while systemic effort is high, or moving a lot during a movement that costs little. Some studies find errors over 50% here.

Nearly meaningless: yoga, pilates, stretching, anything with sustained isometric holds. Little wrist motion, elevated heart rate from postural effort, and the model has no way to reconcile them.

If your training is mostly lifting, the calorie figure on your watch is close to noise.

Myth vs Fact: what people believe about tracker calories

Myth: A chest strap fixes it. Fact: a chest strap makes the heart rate more accurate, which improves the input. The calorie model on top is still a population estimate. Better input, same structural limitation.

Myth: The newer model is much more accurate. Fact: heart rate sensing has improved considerably across generations. Published energy expenditure accuracy has improved far less, because the hard part is not the sensor.

Myth: If it is consistently wrong, it is at least consistently wrong. Fact: partially true and genuinely useful - the error is more stable for you across similar workouts than it is across people. That is exactly why trend comparison works and absolute numbers do not.

Myth: Eating back the calories the watch reports is a reasonable strategy. Fact: this is where the error causes real problems. A 30% overestimate on a 600-calorie reading is 180 calories eaten back that never existed - repeated daily, that fully cancels a deliberate deficit.

Myth: Calories out is the number that matters anyway. Fact: for most people daily non-exercise movement (NEAT) varies far more day to day than workout burn does, and no wrist device captures it well.

So what is the number actually good for?

Plenty, once you stop treating it as currency.

• Week-over-week comparison of yourself. If last week’s runs read 420, 430, 415 and this week they read 380, something changed - fitness, fatigue, heat, or effort. The direction is informative even when the magnitude is not.
• Effort verification. Did today’s session actually match the intensity it felt like? Heart rate zones answer this well, and those are directly measured.
• Consistency and streaks. The behavioral value of a tracker is well supported even where the numbers are not.
• Resting heart rate trend. Arguably the most valuable metric on the device and the one people ignore. A resting heart rate creeping up over several days often precedes illness, poor sleep, or overtraining.
• Sleep duration. Rough on staging, decent on total time.

A reasonable rule: trust the trends, ignore the totals.

What should I use instead if I need real numbers?

Depends on how much precision is worth to you.

Free and surprisingly effective: track intake and body weight for three weeks, then calculate backward. Your actual maintenance calories are whatever intake held your weight steady - that is a measured number specific to you, not an estimate. This beats every device.

Modest cost: a chest strap for accurate heart rate, and use zone time rather than calories as your load metric.

Lab measurement: indirect calorimetry, which measures oxygen consumption and carbon dioxide production directly. Available at some universities and sports medicine clinics for roughly $75-150. This is the only genuinely accurate consumer-accessible measurement of resting metabolic rate.

Not worth it: most consumer at-home metabolic testing gadgets, which have their own validation problems.

The honest answer for most people: you do not need the number. You need consistent effort and a stable intake, and you can verify both without knowing your true burn to the calorie.

FAQ: Tracker Accuracy Questions, Answered

Should I stop wearing my tracker?

No. The behavioral benefit is real - people who track tend to move more. Just move your attention from the calorie tile to resting heart rate, zone time, and step consistency.

Why does my watch give different burns for identical workouts?

Heart rate varies with sleep, hydration, heat, caffeine, and stress. Some of that variance is real physiology, some is sensor noise from fit and skin contact. Wear it snug, one finger-width above the wrist bone, which meaningfully reduces the noise portion.

Does it overestimate or underestimate more often?

Most studies find overestimation is more common, particularly for strength work and low-motion activity. Assume the number is optimistic.

TL;DR:

• Typical calorie error is 20-30%, sometimes far worse - heart rate is accurate, the calorie model on top is not.
• Worst for strength training, yoga, and intervals. Best for steady-state cardio.
• Never eat back tracker calories - a 30% overestimate quietly erases a real deficit.
• Use it for trends: resting heart rate, zone time, week-over-week comparison against yourself.
• Your true maintenance calories come from three weeks of intake plus stable weight, not from a wrist device.

The watch is a decent training log and a poor accountant. Read it that way and it becomes useful again.