6 DEVICES · 7 ACTIVITIES · PEER-REVIEWED

How accurate is your calorie burn?

No wearable measures calories. They estimate energy from heart rate, motion and your profile. Enter the number your device reported and see the likely actual range, using real per-activity research.

Check your number

0

Devices that truly measure calories

27 to 93%

Calorie error across 7 devices · Stanford

~96%

Heart-rate accuracy (what they get right)

52.4%

Calorie error while sitting, the worst of any activity · Stanford

The calculator

Turn your reported burn into a real range.

Pick your device and activity, enter what it showed, and we apply the peer-reviewed error and bias for that exact pairing.

Apple

Fitbit

Garmin

WHOOP

Oura

Samsung

Steady cardio

Running

Walking

Cycling

HIIT

Strength

Swimming

kcal

Reveal my likely burn

Your likely range appears here

Choose a device and activity, enter your reported calories, then reveal the peer-reviewed actual range.

Get the Kygo app

By activity

Accuracy swings wildly by activity.

The same device is decent for steady cardio and hopeless for strength. Only cells with a named study show a number, and every figure names its device model and sample size. Scroll sideways on mobile.

Underestimates Overestimates Variable No peer-reviewed per-activity data † Brand-funded study

† Samsung running is from Ferreira et al. 2026, funded by Samsung, with all authors at Samsung partner Sidia. WHOOP has no published calorie study, so its row is blank. Oura's one lab test (Ring Gen 2, Kristiansson 2023) was 21.1% off against lab gas analysis averaged over eight activities, but it reported error in METs, not calories, so its row is blank too.

Unless marked, every filled cell is a named study and the value is the percent error it published (a MAPE, or the study's typical miss). Cycling and strength come from the newest head-to-head studies, Kostrna 2026 and Yun 2026, which tested four brands on the same people. Where a source publishes a range only, we show the range rather than invent a midpoint. Blank cells mean no per-activity calorie figure exists for that pairing, including Oura for cycling, HIIT, strength and swimming and WHOOP for any activity, none of which have ever been metabolically tested.

Get the free State of Wearable Accuracy report

Brand by brand

How each device gets its number.

Tap any wearable for its method, sensors, best and worst measured activities, and where it breaks down.

Apple Watch

25.95% max treadmill test (Ultra 2), 3rd of four brands head to head

How it calculates

Proprietary ML neural networks trained on metabolic-chamber studies. Combines heart rate, motion, GPS speed, elevation and your profile. Active + Resting calories are summed for the total.

• Sensors Optical HR (green-LED PPG), 3-axis accelerometer, gyroscope, GPS, barometric altimeter.
• Resting rate Resting rate likely from the Harris-Benedict equation (Apple does not disclose it).
• Best measured Walking (~20%)
• Worst measured Strength (~112.2%)
• Tendency Reads high in the newest head-to-head studies: about 14% high on endurance and 112% high on weights (Lee 2026). Sex is not a tested moderator.

Strengths

• Most-studied device, a 56-study meta-analysis (Choe & Kang 2025)
• Reasonable for steady-state cardio and walking
• Heart rate itself is strong: no subgroup in that meta-analysis exceeded the 10% error threshold*

Watch-outs

• Weight training 112.2% off (Series 8, Yun 2026, n=70), and it reads high: +112% in Lee 2026
• Cycling 35.7% off, reading 21.6 kcal high (Series 8, Kostrna 2026, n=58)
• The often-quoted ~28% from the meta-analysis traces to a press release and is pooled bout-level, not a daily total: the paper's own pooled figure is 0.30 kcal/min (limits -2.09 to 2.69)

Read the study View Apple on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

Fitbit

13.97% max treadmill test (Charge 6), 1st of four brands head to head

How it calculates

Heart rate is the primary driver: each minute in an HR zone becomes METs, then calories using your weight. Steps only matter indirectly by raising HR.

• Sensors 3-axis accelerometer, optical HR, altimeter, gyroscope; Charge 6 adds ECG, SpO2, EDA and skin temperature.
• Resting rate A standard metabolic equation (likely Mifflin-St Jeor, undisclosed).
• Best measured Running (12 to 15%)
• Worst measured Walking (~69%)
• Tendency Near-zero average bias, but wide individual swing (-5 to +6 kcal/min). Chevance's own framing: 3 kcal/min over an hour is about 180 kcal, roughly a 40% miss.

Strengths

• Lowest error of four brands in the newest head-to-head studies: 13.97% on a max treadmill test (Charge 6), 31.7% cycling (Sense 2) and 41.7% weight training (Charge 5)
• Closest of four devices on weights in Lee 2026: 145.95 kcal vs a 140.79 kcal criterion, about 4% high
• Bias near zero on all comparisons in a 52-study meta-analysis (k=49: 0.19 kcal/min, SD 2.53, limits -5.32 to 5.70)

Watch-outs

• Walking overestimated by 69% (Charge 2, O'Driscoll 2020, n=59)
• The 31.7% cycling figure holds only after 7 implausible readings were removed; with them the Sense 2 read 128.6 kcal high (Kostrna 2026)
• Remove the low-quality studies from that meta-analysis and the same table gives k=29: -2.77 kcal/min, limits -12.75 to 7.41

Read the study View Fitbit on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

Garmin

14.65% max treadmill test (Forerunner 265S), unplaced on calories

How it calculates

Firstbeat engine: beat-by-beat R-R interval analysis derives respiration rate from HRV, estimates VO2, then converts to METs → calories. Adding respiration lifts accuracy over HR-only methods.

• Sensors PPG optical HR (Elevate Gen 4 red/IR; Gen 5 adds green LEDs), accelerometer, GPS, ANT+/BLE chest-strap compatible.
• Resting rate A standard metabolic equation (undisclosed); nudged up for light daily movement.
• Best measured Running (~22%)
• Worst measured Strength (~114.9%)
• Tendency Reads high in the newest studies: +30.5% on endurance and +116.4% on weights (Lee 2026). A 2020 lab review on older hardware found underestimation.

Strengths

• Most sophisticated engine (Firstbeat HRV modeling)
• Treadmill running holds up: 14.65% on a max treadmill test (Forerunner 265S, Perez 2025) and 19.1% in de Leon 2026 (Vivoactive 4, n=18, unfunded)
• A paired chest strap sharply improves accuracy

Watch-outs

• Cycling 77.6% off, reading 68.6 kcal high (Forerunner 955, Kostrna 2026)
• Weight training 114.9% off (Vivosmart 5, Yun 2026, n=70), and Lee 2026 found it reading +116.4% high on weights
• Fuller 2020 (lab only, 51 non-independent comparisons on 2013 to 2019 hardware, one author employed by Garmin during publication) found underestimation, the opposite of the newer studies

Read the study View Garmin on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

WHOOP

No study no WHOOP calorie validation has been published

How it calculates

ACSM metabolic equations extended by Keytel et al. 2005: Calories = BMR + f(heart rate), activating once HR rises above your resting baseline. Recovery status shifts the estimate.

• Sensors Advanced PPG (MAX86171 on 4.0), accelerometer, skin temperature, SpO2; 5.0 adds ~26 Hz sampling + respiratory rate.
• Resting rate Age, sex, height, weight; 5.0 adds a 30-day personalized calibration window.
• Tendency Unknown. With no calorie study there is no measured direction. The estimate is recovery-coupled, so the same workout can read differently by day.

Strengths

• Built for relative strain and recovery trends
• The HR and HRV signal itself is well-validated by Bellenger et al. 2021, which validated HR and HRV only, never calories
• 5.0 adds a 30-day personalized calibration window

Watch-outs

• No calorie study of any WHOOP model has been published, so this tool publishes no WHOOP calorie figure
• Calorie percentages that circulate online for WHOOP trace to no primary publication (no authors, DOI or PubMed record)
• No GPS and no screen, so workouts rely on heart rate alone

Read the study View WHOOP on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

Oura Ring

21.1% lab average of 8 activities (Ring Gen 2, 2023), not ranked

How it calculates

BMR + activity calories via METs across the day (resets at 4 AM). A Nov 2024 update folded in HR intensity, cutting median active-calorie error by 53%. Recognizes 40+ activity types.

• Sensors 18-path multi-wavelength PPG (red/IR/green LEDs), 3D accelerometer, 2 precision thermistors.
• Resting rate A standard metabolic equation using age, sex, height and weight.
• Tendency Underestimates as effort rises: 0.4 MET low on average in the one lab test, and 3.5 MET low when running at 15 km/h (Ring Gen 2, Kristiansson 2023).

Strengths

• Built for daily activity and recovery trends
• Finger PPG gives a cleaner resting signal than the wrist
• A Nov 2024 update folded heart rate into activity calories (an Oura figure, not a study)

Watch-outs

• 21.1% off against lab gas analysis on average, with the miss growing as running speed rose (Ring Gen 2, Kristiansson 2023)
• Its only daily-total comparison used a wrist accelerometer as the reference, not lab gas analysis, so this tool does not use it
• No GPS; poor for cycling/elliptical (no hand motion)

Read the study View Oura on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

Samsung Galaxy Watch

10 to 13% running (Galaxy Watch 6, 7)†, 2nd of four brands head to head

How it calculates

HR-VO2 relationships combined with accelerometer and GPS; MET tables applied to recognized activities. Watch 7 adds a BIA body-composition sensor whose role in the calorie model is undocumented.

• Sensors Optical HR (PPG), 3-axis accelerometer, gyroscope, barometer, GPS; Watch 7 adds the BioActive sensor (ECG, BIA body composition).
• Resting rate A standard metabolic equation (undisclosed, Harris-Benedict or Mifflin-St Jeor).
• Best measured Running (10 to 13%)
• Worst measured Strength (~84.5%)
• Tendency Close on running and walking, but reads high on cycling and weights (+83% on weights, Lee 2026).

Strengths

• Lowest error of five brands on a max treadmill test: 9.42% (Galaxy Watch 7, Perez 2025, conference abstract)
• Running 10 to 13% (Galaxy Watch 6, 7, Ferreira 2026, n=148)†
• Walking within 0.5% of the criterion total (Galaxy Watch 5, Oh 2024, 27 men, 30-minute walk)

Watch-outs

• Cycling 66.5% off, reading 56.8 kcal high (Galaxy Watch 5, Kostrna 2026)
• Weight training 84.5% off (Galaxy Watch 5, Yun 2026, n=70), reading 83% high in Lee 2026
• † The running study was Samsung-funded and all its authors work at Samsung partner Sidia

Read the study View Samsung on Amazon

Affiliate link. We may earn a commission at no extra cost to you.

The "View on Amazon" links above are affiliate links. As an Amazon Associate, Kygo Health earns from qualifying purchases, at no extra cost to you.

Hidden variables

Factors that skew every device.

Even the best wearable can't see these, and each one shifts the calorie estimate.

Body composition / BMI

High impact

Algorithms are trained on average builds. Very lean, muscular or higher-BMI bodies skew the HR-to-calorie conversion, and Fitbit is specifically less accurate at higher BMI.

Medications (beta-blockers, stimulants)

High impact

Beta-blockers cap heart rate and cause big underestimation; stimulants raise resting HR and inflate the count. Devices can't detect either.

Tattoos

High impact

Dense or dark ink over the sensor blocks light transmission and can make wrist PPG readings unreliable.

Device fit & placement

High impact

Loose bands, wrist hair and low placement all add motion artifact. The band should sit snug, 1 to 2 finger-widths above the wrist bone.

Skin tone

Moderate

LED optical sensors read darker pigmentation less consistently. Garmin Elevate Gen 5 and Apple Series 9+ improved this, but a Series 9 study still found tracking inconsistent across pigmentation groups.

Sex

Moderate

Shcherbina et al. 2017 found sex a significant predictor of heart rate error across seven devices, with error rates for men running about 4% higher than for women. That is heart rate, not calories, and it is not specific to any one brand. No calorie meta-analysis has tested sex as a moderator.

Age

Moderate

Optical signal quality drops with skin changes, and the 220−age max-HR assumption drifts further off with age.

Caffeine & hormonal state

Moderate

Caffeine lifts heart rate independent of effort; menstrual cycle, thyroid and other hormonal shifts move metabolic rate in ways no wearable can see.

The bottom line

Wearables get heart rate right (~96%) but the leap from heart rate to calories carries 15 to 40%+ real-world error, up to 93% in the Stanford study. Steady cardio is the safe zone; strength training and cycling are the worst regardless of brand, and pooled across the Stanford devices the error was actually highest at rest (52.4%), not during exercise. Use the number for week-over-week trends and relative effort, never as an exact figure to eat back.

Keep exploring

Related tools.

More free, evidence-based tools to get the most out of your wearable.

Daily step counts

Step Count Accuracy

Which wearable counts steps most accurately, ranked by 20+ peer-reviewed studies.

Wearables · 9 devices

Open

What moves accuracy

Wearable Accuracy: What Actually Affects It

Tattoos, hairy arms, cold hands, strap tightness, which wrist. What the studies found and the free fixes that work tonight.

Wearables · 26 answers

Open

Fitbit Air vs WHOOP

Fitbit Air vs WHOOP

Heart-rate, sleep and calorie accuracy with numbers attached, plus the 3-year cost.

Wearables · head-to-head

Open

Want every brand on every metric in one place? Open the wearable accuracy hub

FAQ

Common questions.

How accurate is Apple Watch calorie burn?

Apple Watch placed third of four brands head to head: 25.95% off on a max treadmill test (Ultra 2, Perez 2025, conference abstract), 35.7% on cycling (Series 8, Kostrna 2026) and 112.2% on weight training (Series 8, Yun 2026), where it reads high. It reads best for walking (~20%) and running (~24%) in Le 2022. The often-quoted ~28% from a 56-study meta-analysis (Choe and Kang 2025) traces to a press release and is pooled across bouts; the paper's own figure is 0.30 kcal/min (limits -2.09 to 2.69).

Which wearable is most accurate for calories?

Fitbit, then Samsung, then Apple Watch, but no brand came within 9% on the max treadmill test. In head-to-head lab studies, Fitbit was 13.97% off on a max treadmill test (Charge 6), 31.7% on cycling (Sense 2) and 41.7% on weights (Charge 5). Samsung was 9.42% on the treadmill (Galaxy Watch 7) but 66.5% on cycling and 84.5% on weights (Galaxy Watch 5). Apple Watch was 25.95%, 35.7% and 112.2%; Garmin 14.65%, 77.6% and 114.9%. WHOOP has no calorie study, and Oura's one lab test (Ring Gen 2) was 21.1% off averaged over eight activities, a different protocol that is not ranked. Fuller 2020 found no brand within 3% of the criterion more than 13% of the time. In the Stanford seven-device study (Shcherbina 2017), calorie error was lowest for walking (31.8%) and running (31.0%) and highest for sitting (52.4%), so the worst errors happen at rest.

Why does my wearable overestimate or underestimate calories?

Wearables do not measure calories directly. They infer them from heart rate, motion and your profile, so the conversion carries the error. Sex, skin tone, body composition, medications (beta-blockers, stimulants), tattoos and band fit all shift the result. Static, low-motion work like strength training is the hardest case for every brand.

Is the calorie number on my watch good enough for a diet?

Not for precise nutrition math. Real-world calorie error runs 15 to 40%+, and the Stanford 2017 study found some devices off by up to 93%. Use the number for week-over-week trends and relative effort, not as an exact figure to eat back.

How accurate is Oura, WHOOP or Garmin for exercise calories?

Oura has one lab test against gas analysis: Ring Gen 2 was 21.1% off averaged over eight activities from sitting to running (Kristiansson 2023, n=28), reading lower as running speed rose. It was measured in METs on a different protocol, so it is not ranked against the head-to-head studies. No calorie study of any WHOOP model has been published; Bellenger 2021 validated WHOOP's heart rate and HRV only, never its calories. Garmin is close on treadmill running, 14.65% on a max treadmill test (Forerunner 265S, Perez 2025) and 19.1% in de Leon 2026 (Vivoactive 4, n=18), but reads high on cycling (77.6%, Forerunner 955, Kostrna 2026) and weight training (114.9%, Vivosmart 5, Yun 2026).

How does this calculator work?

Pick your device and activity, enter the calories it reported, and the tool applies the published per-activity error and bias direction for that exact combination to show your likely actual range. Every figure names its device model and sample size. With no direct per-activity study the result is flagged as an estimate, a brand-funded study is marked with a dagger, and a source that publishes only a range shows the range, not an invented midpoint.

Sources

Every claim, traceable.

Each checked against the primary record (PubMed / PMC / journal / manufacturer), with the device model and sample size named. Re-verified September 2026.

Anchor · multi-device

Shcherbina et al. 2017: Stanford 7-device study

J Personalized Medicine · n=60 · 27 to 93% energy-expenditure error. Lowest for walking (31.8%) and running (31.0%), highest for sitting (52.4%). Apple had the most favourable overall error profile, PulseOn the least.

Apple / Garmin / Huawei ⚠️ funded

Le et al. 2022, Front Physiol 13:995575: walking and running

⚠️ Apple Watch Series 6, Garmin and a Huawei Watch GT 2e vs COSMED K5, n=20. Funded by Huawei Terminal Co. LTD, whose device wins both conditions (9.9% and 11.9%), while the paper's conflict statement claims no commercial relationship.

Fitbit / Apple / Garmin

JMIR mHealth 2017: Charge HR comparison

n=62 · error rises with intensity

Gold standard

Murakami et al. 2019: 12-device DLW validation

JMIR mHealth · only 2 of 12 acceptable free-living

Multi-device review

Fuller et al. 2020, JMIR mHealth: wrist-wearable review

⚠️ Laboratory only · 51 non-independent comparisons on 2013 to 2019 hardware · one author was employed by Garmin during the publication process. Headline finding: no brand was within 3% of criterion more than 13% of the time. Underestimation in 69% of observations applies to the lab data only.

⚠️ Funded · brand attribution

Lee et al. 2026, Sensors 26(8):2526: four-device energy expenditure

⚠️ n=62 men · SOLUM-funded with an employed co-author, and the abstract contradicts its own tables, so cite Tables 1 and 2 only. Reads high on endurance and weights: Fitbit +21% and +4% (145.95 vs 140.79 kcal), Samsung +9% and +83%, Apple Watch +14% and +112%, Garmin +30.5% and +116.4%.

Head to head · conference abstract

Perez et al. 2025, IJES Conference Proceedings 14(5):102: max treadmill test

About 20 adults vs TrueOne 2400 · conference abstract, not peer reviewed. MAPE: Galaxy Watch 7 9.42%, Fitbit Charge 6 13.97%, Garmin Forerunner 265S 14.65%, Polar Vantage M3 21.10%, Apple Watch Ultra 2 25.95%.

Head to head · cycling

Kostrna et al. 2026, PLOS ONE: four watches on a 10-minute cycle

n=58 Hispanic adults vs COSMED K5 · NSF grant, no device conflict. Typical miss: Fitbit Sense 2 31.7% (after 7 implausible readings removed), Apple Watch Series 8 35.7%, Galaxy Watch 5 66.5%, Garmin Forerunner 955 77.6%.

Head to head · weight training

Yun et al. 2026, Bioengineering: four watches in weight training

n=70 men, 34-minute session vs TrueOne 2400 · no funding declared. MAPE: Fitbit Charge 5 41.7%, Galaxy Watch 5 84.5%, Apple Watch Series 8 112.2%, Garmin Vivosmart 5 114.9%.

Why brand attribution matters

Parak et al. 2017, JMIR mHealth: Firstbeat modeling in a PulseOn

Often mis-cited as a Garmin result. No Garmin device was tested; the device is a PulseOn and two authors were PulseOn employees. Stanford ranked PulseOn worst of seven devices for energy expenditure (92.6%).

Anchor · Apple

Choe & Kang 2025: Apple Watch meta-analysis

Physiological Measurement · 56 studies · pooled 0.30 kcal/min (limits -2.09 to 2.69). The 27.96% figure is not in the abstract and traces to the university press release. Moderators tested: age, health status, device series, activity intensity, activity type. Sex is not among them.

Apple

Sun et al. 2023, J Sci Med Sport: Apple Watch 6 running

n=11 · rest and running protocols only, no resistance protocol. Treadmill 447.91 vs a 367.14 kcal criterion, bias +80.77. Sun defines MAPE as (criterion - estimate)/criterion, so a positive signed value there means underestimation.

Apple

MDPI Sensors 2024: Series 9 skin pigmentation

Tracking inconsistent across pigmentation groups

Apple

Apple: Heart Rate & Calorimetry white paper (2024)

Manufacturer method documentation

Anchor · Fitbit

O'Driscoll et al. 2020, Health and Technology 10(3):637-648

n=59 · Charge 2 vs a Vyntus CPX metabolic cart. Walk at 4 km/h: 69% MAPE (7.10 vs 4.27 kcal/min). Running: 15% flat, 12% incline. No manufacturer funding.

Fitbit

Chevance et al. 2022, JMIR mHealth: Fitbit meta-analysis

52 studies · Table 2, all comparisons (k=49): bias 0.19 kcal/min (SD 2.53), limits -5.32 to 5.70. Low-quality studies removed (k=29): bias -2.77 (SD 4.12), limits -12.75 to 7.41. The authors note 3 kcal/min over an hour is about 180 kcal, roughly a 40% miss.

Fitbit

Chowdhury et al. 2017, PLoS ONE 12(2):e0171720

n=30 · Charge HR vs a calibrated Actiheart over ~36 hours, daily bias -405 ± 944 kcal/day. An accelerometer reference, not calorimetry, so it is no longer used as a headline here.

Fitbit

Fitbit: how calorie burn is calculated

Manufacturer method documentation

Garmin

Boudreaux et al. 2018, Med Sci Sports Exerc 50(3):624-633

PMID 29189666 · n=50 · Vivosmart HR 57.02% MAPE for resistance. Its own conclusion: no device was valid for energy expenditure during cycling or resistance exercise. Also the true source of a resistance figure often misattributed to Apple, which belongs to the Polar A360.

Garmin

de Leon et al. 2026, MDPI Applied Sciences 16(3):1286

Vivoactive 4 treadmill · n=18 · 19.1% MAPE · no funding, no declared conflict. Published 27 January 2026.

Garmin

Firstbeat: energy expenditure white paper

Manufacturer method documentation

WHOOP

Bellenger et al. 2021: HR/HRV validation

Sensors · validates heart rate and HRV only, never calories

WHOOP

Keytel et al. 2005: HR-to-EE equation

J Sports Sciences · the basis of WHOOP's algorithm

WHOOP

WHOOP: how calories are calculated

Manufacturer method documentation. The page, updated 9 May 2025, now states that WHOOP prioritizes accuracy over overestimation.

Oura

Kristiansson et al. 2023: Oura energy expenditure validation

BMC Med Res Methodology 23:50, corrected 9 September 2023 · Oura Ring Gen 2 · n=28 healthy adults vs Oxycon Pro lab gas analysis: 21.1% MAPE in METs averaged over 8 activities, rising with running speed. Its 13.0% free-living figure used a wrist accelerometer as the reference, so it is not used here. University and foundation funded, no competing interests.

Oura

Oura: activity improvements (Nov 2024)

HR integration cut median active-calorie error 53%

Samsung ⚠️ funded

Ferreira et al. 2026, JMIR Formative Research: Galaxy Watch 6 and 7

⚠️ n=148 adults, 27-minute intermittent running vs COSMED K5. MAPE 10.10% (Galaxy Watch 6) and 12.55% (Galaxy Watch 7). Samsung-funded; all authors at Samsung partner Sidia.

Samsung

Oh et al. 2024, Asian J Kinesiology: Galaxy Watch 5 walking

n=27 men, 30-minute walk vs COSMED Quark CPET. 180.3 vs 179.5 kcal, 0.5% high; r 0.95.

Show all 28 sources

From the blog

Show fewer sources

Keep reading.

The long-form, evidence-based articles behind this tool.

Wearables & Data

How Accurate Is Your Wearable's Calorie Burn? 5 Devices Compared by Research

Every major wearable gets calorie burn wrong by 15-40%. How Apple Watch, Fitbit, Garmin, WHOOP and Oura each calculate it, and where they fail.

10 min read

Read

Wearables & Data

How Accurate Is Apple Watch Calorie Burn? (2026 Research)

Apple Watch calorie burn is off by roughly 20-40% depending on the activity. The peer-reviewed research, broken down by workout type.

8 min read

Read

Wearables & Data

How Accurate Is Fitbit Calorie Burn? (2026 Research)

Fitbit's average bias is near zero but individual readings swing widely. Running is its best case, walking its worst.

7 min read

Read

This calculator is for educational purposes only and is not medical or nutrition advice. Wearable calorie estimates vary by individual, environment and algorithm updates. Always consult a qualified professional before making health decisions based on wearable data.

Data from peer-reviewed studies, meta-analyses and manufacturer documentation. Last updated October 2026.

As an Amazon Associate, Kygo Health earns from qualifying purchases. Product links on this page are affiliate links; we may earn a commission at no extra cost to you.