How Accurate Are Garmin Calories?

What Garmin Is Actually Measuring

No wrist device measures calories. Measuring energy expenditure requires capturing the gases you breathe (indirect calorimetry) or tracking labelled water through your body over weeks. What Garmin measures is heart rate, motion, and in some cases barometric altitude and GPS speed; calories are a model's best inference from those signals. Understanding the model tells you when to trust it.

How the Garmin Calorie Model Works

Garmin's physiology analytics come from Firstbeat, which Garmin acquired in 2020. The calorie estimate has two components that appear separately in Garmin Connect.

Resting calories

Resting calories represent basal metabolic rate over 24 hours, derived from your profile: age, sex, height and weight. This is the same class of estimate as the Mifflin-St Jeor equation, and it has the same weakness — a prediction equation is right on average across a population but can be off by 10% or more for an individual, especially at very high or very low body-fat levels. It is also only as accurate as the weight you entered.

Active calories

Active calories are estimated from heart rate during recorded activities and from movement the rest of the day. The heart-rate method rests on a well-established relationship: for a given person, oxygen consumption (and therefore energy production) rises roughly linearly with heart rate across the aerobic range. Firstbeat personalises that line using your estimated VO2 max, your maximum heart rate and your resting heart rate, and it adds a component for the elevated metabolism that persists after hard efforts (the excess post-exercise oxygen consumption, or EPOC, that Garmin surfaces as Training Effect). Outside workouts, the accelerometer translates steps and movement intensity into energy using your body mass.

Typical Error Bands

Independent laboratory comparisons of consumer wearables against indirect calorimetry consistently reach the same headline: heart rate is measured well, energy expenditure is the weakest metric for every brand. A widely cited 2017 university study of seven wrist devices found that none estimated energy expenditure within 20% on average during lab activities, and the worst was off by more than 90% — Garmin was not among the tested devices, but the finding reflects the limits of the method rather than any one brand. Firstbeat's own published validation work reports smaller errors for its heart-rate-based approach during steady aerobic exercise when VO2 max and maximum heart rate are known accurately, which is exactly the condition in which the model is strongest.

Putting these together, a practical set of expectations for a Garmin worn correctly with a current profile:

When Garmin Over-Estimates

• Strength training. Heart rate spikes from bracing and breath-holding, not oxygen demand. A 45-minute lift that Garmin scores at 450 calories often costs 250–300 by calorimetry.
• Heat and dehydration. Cardiovascular drift pushes heart rate up 5–15 beats at the same pace late in a hot session; the model reads that as more work.
• Caffeine, stress, poor sleep. All raise heart rate at a given output. A pre-workout with 300 mg caffeine can add a noticeable margin to an easy run's estimate.
• An out-of-date VO2 max. If your fitness has improved but the watch has not updated its estimate (common if you only do non-GPS activities), the model assumes each heartbeat buys less oxygen than it now does — a subtle overestimate. Conversely a stale, too-high VO2 max after time off underestimates.
• Wrist profile weight too high. Every component scales with mass.

When Garmin Under-Estimates

• Optical heart-rate dropouts. Cold hands, tattoos, a loose strap or wrist flexion (cycling, rowing, kettlebells) cause the sensor to lose lock and fall back on motion. Pairing a chest strap fixes most of this.
• Arm-static work. Pushing a stroller, carrying groceries, walking with hands in pockets: steps go uncounted and heart rate outside a recorded activity is sampled less aggressively.
• Very lean, muscular people. Resting metabolism is higher per kilogram than the population equation assumes.
• Watch off the wrist. Charging during your morning routine or leaving it off in the gym shower is a silent hole in the day's total.

How to Calibrate: The 2-Week Check

The most reliable calibration instrument you own is a bathroom scale. Body weight over two weeks integrates every error in the model into a single correction factor. Here is the protocol:

1. Update your Garmin profile with current weight and confirm your maximum heart rate setting is realistic (if you have never seen 195 in a race, do not leave it at 220 minus age).
2. Eat to Garmin's Total Calories (resting plus active) each day for 14 days, tracking intake as accurately as you can. Do not add exercise calories on top.
3. Weigh yourself every morning after the bathroom, before food or water, and average days 1–7 and days 8–14.
4. Compute the correction. Roughly 7,700 calories corresponds to one kilogram of body fat. If you gained 0.5 kg between the two weekly averages, Garmin over-reported by about 3,850 calories over 14 days — around 275 per day. If you lost 0.5 kg, it under-reported by the same amount.
5. Apply it. Subtract (or add) that daily correction to Garmin's total going forward, and re-check every 8–12 weeks or whenever your training changes substantially.

How to Eat to Garmin Numbers Safely

Use the 7-day average, not today's total

Active calories can swing from 250 on a rest day to 2,000 after a long ride. Matching intake to each day precisely is unnecessary and usually leads to under-eating on rest days when your body is doing most of its repair. Anchor to the weekly average, then bias carbohydrate up on hard days and down on easy ones. Our Garmin macro calculator is built around that average.

Discount the categories that inflate

If your training is mostly lifting or HIIT, mentally haircut Garmin's active calories by 20–25% before planning a deficit. If it is mostly steady running with a chest strap, take the number at face value.

Protect protein regardless of the calorie error

Protein needs scale with body weight, not with Garmin's opinion of your expenditure. Fix protein at 1.6–2.2 g/kg first and let the calorie error land on carbohydrate and fat, where it does far less harm to lean mass.

Keep the deficit modest

Because the estimate can be 300 calories high, a planned 500-calorie deficit may really be 200 — or, if it is 300 low, a punishing 800. A 400–500 planned deficit with a two-week check is the safe compromise. Never build a diet on a Garmin number that has not been tested against the scale.

Watch Garmin's own warning lights

Chronic under-fueling shows up in Garmin data before it shows on the scale: morning Body Battery stops reaching the 80s, HRV Status drifts to "low" or "unbalanced," resting heart rate creeps up 3–5 beats, and Training Readiness sits low despite easy days. If you see that cluster while losing weight, add 200–300 calories of carbohydrate and reassess in a week.

Let the Model Learn You Automatically

Everything above is what Plait does behind the scenes for Garmin users. When you connect Garmin, your daily calories, Body Battery, sleep, stress and workouts flow into Plait automatically. Plait anchors your targets to your weekly measured expenditure, holds protein at a body-weight-based floor, biases carbohydrate by your morning Body Battery, and builds the meals to match — so the Garmin number becomes a starting point that gets more accurate for you every week, rather than a guess you have to trust.

Further Reading