Watch Calorie Accuracy: Why Garmin, COROS, and Apple Watch Overestimate Burn

Your watch finished the run at 812 calories. The MET estimate said ~640. The treadmill said 590. Which number do you eat back? For most athletes wearing Garmin, COROS, or Apple Watch, the device total is the highest — and the least trustworthy for precise nutrition. Wearables are excellent for consistency and trends. They are mediocre laboratory instruments.

This guide explains where Firstbeat-style models and optical heart-rate sensors introduce error, when ±15–25% misses are common, and how to use a MET-based calculator as an independent check. For the underlying running math, see How Many Calories Does Running Burn? and the Running Calories Calculator.

What Your Watch Is Actually Estimating

Consumer watches do not measure calories directly. They estimate energy expenditure from a stack of inputs:

• User profile (age, sex, weight, height — sometimes VO2 max estimate)
• Movement (GPS speed/pace, accelerometer, occasionally power)
• Heart rate (optical wrist PPG or paired chest strap)
• Proprietary models (Garmin Firstbeat analytics, Apple energy algorithms, COROS training models)

Each layer adds uncertainty. Weight wrong by 3 kg shifts the baseline. Optical HR wrong by 10–15 bpm during intervals shifts the intensity assumption. The model then confidently displays a single integer that looks like a measurement.

Also confirm active vs total calories. Garmin Connect often shows activity calories that include a basal component for the workout duration. Apple separates Active Energy from Resting Energy more clearly in Health, but workout tiles still confuse users. COROS typically emphasizes active burn but still depends on HR quality. Logging "total" as if it were "active" is a common way to invent a 150–250 kcal surplus snack you did not earn.

The MET Baseline You Should Compare Against

Calories per minute = (MET × body weight kg × 3.5) / 200

Worked example — 70 kg runner, 50 minutes at easy pace (~9.8 MET):

(9.8 × 70 × 3.5) / 200 = 12.005 kcal/min × 50 ≈ 600 kcal.

Distance shortcut (≈1 kcal/kg/km): 70 kg × ~8.2 km ≈ 574 kcal. A sensible planning range is about 570–620 kcal.

If Garmin reports 780 active calories for that same flat easy run with stable pace, you are looking at roughly +25–30% versus MET/distance anchors. That magnitude of overestimate is common enough that PaceWOD recommends treating watch calories as an upper bound until proven otherwise against multi-week weight trends.

Run the same session through the Running Calories Calculator and keep a simple note: watch, MET, midpoint. After three weeks, the midpoint that matches your scale trend is your personal correction factor.

Firstbeat and HR-Driven Models: Strengths and Failure Modes

Garmin's Firstbeat-derived analytics improved on crude speed-only formulas by incorporating heart rate and personalized fitness estimates. When HR is accurate and the effort is continuous (steady run, bike, row), estimates often land closer to reality than older pedometer-era guesses.

Failure modes still matter:

• HR overstatement → calorie overstatement: If wrist HR reads 155 when true HR is 140 on an "easy" run, the model thinks you are working harder than you are.
• Fitness profile lag: An inflated VO2 max estimate can alter expected HR–pace relationships and distort calorie math in either direction.
• Stop-and-go sports: Trails with hiking segments, city runs with lights, and CrossFit-style sessions keep HR elevated while mechanical work drops. Models that weight HR heavily keep the burn high during those gaps.
• Heat and dehydration: Cardiac drift raises HR at the same pace. Some of that extra cost is real (thermoregulation); some is model overreaction. Hot-weather runs often look "more expensive" on the watch than MET tables predict for the pace alone.

COROS uses its own stack but shares the same vulnerability: garbage HR in, optimistic calories out. Apple Watch Move calories are notoriously sticky for some users — rings close easily when HR is elevated even during light activity — which trains the wrong mental model for deficit planning.

Optical Heart Rate: The Quiet Culprit

Wrist PPG (photoplethysmography) estimates blood-volume changes with LEDs. It works acceptably for steady Zone 2 on a well-fitted watch. It degrades when:

• Cadence and arm motion create motion artifact (especially intervals and trail technical sections)
• The watch sits loose, too far up the wrist, or over prominent tendons
• Cold weather reduces peripheral perfusion
• Grip-intensive work (pull-ups, farmers carries, barbell cycling) distorts the optical signal
• Dark tattoos or certain skin/optical combinations weaken signal quality

When optical HR spikes or plateaus incorrectly, calorie models that depend on it follow. A chest strap (ANT+/Bluetooth) dramatically improves HR fidelity for hard sessions and usually pulls calorie estimates closer to MET expectations — though the proprietary model can still bias high. For brand-specific HR behavior, see Garmin vs COROS heart rate accuracy.

When Watches Are Typically ±15–25% Off

Expect larger errors in these situations:

• Interval workouts: Optical lag under-reads surges and over-reads recoveries; net calorie error of 15–25% is common.
• Strength and CrossFit: Rest intervals with elevated HR inflate totals; 20–40% overestimates versus MET session averages are not rare.
• Trail running with hiking: HR stays high on climbs while pace collapses; models may not correctly blend hiking vs running economy.
• Indoor runs without treadmill speed sensors: Wrist-based distance/pace error compounds calorie error.
• Wrong user weight: Athletes who lose 4–5 kg but never update the profile keep paying the "old weight" calorie tax — or the reverse if weight is under-entered.

Steady outdoor Zone 2 with a chest strap and correct body-weight profile is the best-case scenario. Even then, treat ±10% as normal noise, not failure.

Brand Tendencies Without Mythology

Garmin

Often strong for continuous endurance when HR is clean. Watch for total-vs-active confusion in Connect and for Firstbeat sensitivity to HR drift. Running Dynamics and power (if you use a Stryd or similar) can improve load metrics but do not magically create lab-grade calories.

COROS

Clean training UI and solid GPS on many models. Calorie quality still tracks optical HR quality. Compare against MET on easy runs; if you train by power or pace more than HR, do not assume the calorie field is your best nutrition input.

Apple Watch

Excellent lifestyle activity capture; workout calories for structured running vary. Some runners see optimistic Active Energy on easy days that still "feel hard" in heat. Always cross-check hard sessions against pace–MET math before eating back Move ring calories.

A Practical Hierarchy for Nutrition Decisions

1. Primary: Multi-week weight, waist, and performance trends under a planned intake
2. Secondary: MET or distance-based estimates from the Calories Calculator and running calorie guide
3. Tertiary: Watch active calories, same device, weekly sums, with a personal correction factor
4. Ignore for precision: Social screenshots, ring-closing bonuses, and "total" calories mistaken for active

If you are in a cut, default to the lower end of the MET–watch range when deciding whether to add a snack. Appetite after hard sessions already pushes intake up; you rarely need the watch's optimistic bias on top.

Worked Comparison: Same Runner, Three Tools

Profile: 70 kg, 10 km road run in 55 minutes (~10.9 km/h), cool weather, flat route.

• Distance rule: 70 × 10 = 700 kcal
• MET (~10.5 for this pace): (10.5 × 70 × 3.5 / 200) × 55 ≈ 707 kcal
• Hypothetical watch (optical HR reading ~8–10 bpm high in the second half): 820–880 kcal

Planning number: ~700 kcal. If this runner ate an extra 200 kcal bar "because Garmin said 880," that is most of a day's intended 300–500 kcal deficit gone in one decision. Over a week of similar biases, fat loss stalls while training "looks" perfect in the app.

How to Make Wearables Useful Anyway

• Update body weight monthly (or whenever you change 2+ kg)
• Prefer chest-strap HR for key workouts if calories inform eating
• Use one ecosystem for trends; do not hop between Apple and Garmin totals
• Compare weekly active calories to weekly MET sums, not single runs
• Recalibrate expectations after heat waves, altitude trips, or illness

Watches remain outstanding for pace, distance, HR zones, and adherence. Calorie fields are decision-support, not receipts. Pair them with MET math and honest feedback from the scale and your sessions.

FAQ

Are Garmin calories more accurate than Apple Watch?

For steady outdoor running with good HR data, Garmin's endurance-focused models are often closer to MET estimates than Apple's lifestyle-oriented Active Energy. Neither is consistently lab-accurate. Validate whichever you wear against a MET calculator and your weight trend.

Will a chest strap fix calorie overestimation?

It usually reduces error by fixing the HR input, especially for intervals and cold weather. It does not eliminate proprietary model bias or active-vs-total confusion. Still cross-check hard days with MET math.

Why does my easy run show almost as many calories as my tempo run?

Duration often dominates. A 70-minute easy run at 9.8 MET can outburn a 35-minute tempo at 12 MET for the same athlete. Also check whether optical HR failed to rise appropriately in the tempo — under-read hard efforts and over-read easy efforts both distort comparisons.

Should I eat back watch calories during a cut?

Generally no — or only a fraction, using a haircut of 15–25% when the watch exceeds MET estimates. Plan meals around a weekly deficit and training quality. Use the Running Calories Calculator as the independent check, not the ring that wants to close.