TL;DR
- Average error is small: good watches stay within about 1-5 bpm of ECG on average in lab tests (Apple Watch Series 6 and Garmin Forerunner 945: 1.2-2.3 bpm; Galaxy Watch 6: 1-3 bpm).[1][4]
- Single readings are not: limits of agreement of about ±15 bpm are typical, so any one number on the wrist can be well off.[4][7]
- Device and activity matter more than intensity alone: in a 2026 ten-device study some wearables averaged 9-14 bpm off at walking pace, and stop-start movement raised errors.[3]
- A chest strap stays the reference: the Polar H10 ($104.95) is what the validation studies themselves use as the criterion.[4][5][6]
How accurate is wrist heart rate? On average, good watches read within about 1-5 bpm of an ECG chest strap during steady exercise, but individual readings can miss by 15 bpm or more, and the error depends heavily on the device and the activity.[1][3][4] This page collects the peer-reviewed validation studies behind those numbers, each compared against ECG or a validated ECG strap.
Why do wrist sensors and chest straps disagree?
A chest strap places electrodes against the skin and measures the heart's electrical signal directly, the same signal a clinical ECG reads, so it captures each beat as it happens.[6] A wrist or arm sensor uses photoplethysmography (PPG): it shines light into the skin and infers heart rate from changes in blood flow. That indirect signal is disturbed by arm movement, a loose band and poor blood flow at the wrist, and devices smooth it over several seconds, which is why wrist readings lag sudden changes.
What do the validation studies report?
| Study | Device (where worn) | Reference | Conditions | Result |
|---|---|---|---|---|
| Ho 2022[1] | Apple Watch Series 6 (wrist) | ECG | Maximal treadmill test | MAE 1.16-1.48 bpm; MAPE under 1% |
| Ho 2022[1] | Garmin Forerunner 945 (wrist) | ECG | Maximal treadmill test | MAE 1.35-2.25 bpm; MAPE 1.16-1.39% |
| Inoue 2026[4] | Galaxy Watch 6 (wrist) | Polar H10 | Maximal treadmill test, 55 adults | Median error 1-3 bpm; bias -2.67 bpm; limits of agreement -16.9 to +11.6 bpm |
| Sartor 2018[7] | Wrist optical monitor | ECG chest strap | Walking, running, cycling, gym, 199 adults | MAE 3 bpm or less; limits of agreement within ±15 bpm |
| Gielen 2026[3] | Fitbit Charge 6 (wrist) | ECG chest strap | Rest, stress task, steady and stop-start walking | Median MAE 4.5 bpm; MAPE 5.5% |
| Gielen 2026[3] | Apple Watch SE (wrist) | ECG chest strap | Same | Median MAE 5.0 bpm; MAPE 7.3% |
| Gielen 2026[3] | Polar Ignite 3, Polar Pacer, Oura Gen 3 | ECG chest strap | Same | Median MAE 9.5-11.0 bpm; MAPE 11-15% |
| Schweizer 2025[5] | Polar optical sensor, wrist vs upper arm | Polar H10 | Sedentary to vigorous activity | Wrist MAE 6.41 bpm; upper arm 1.43 bpm |
| Gillinov 2017[2] | Polar H7 strap vs four watches and an armband | ECG | Treadmill, bike, elliptical, 50 adults | Strap concordance 0.996; Apple Watch 0.92; Garmin Forerunner 235 0.81; Fitbit Blaze 0.67 |
Is wrist heart rate accurate for running?
For steady running, mostly yes. Gillinov 2017 found treadmill running was where wrist devices did best (concordance 0.88-0.93 for all but one device),[2] and the Apple Watch and Garmin in Ho 2022 stayed within about 1-2 bpm of ECG through a maximal treadmill test.[1] The newer data also contradict the idea that error simply rises with intensity: in Inoue 2026 the Galaxy Watch's percentage error fell from 2.9% at 50-60% of maximum heart rate to under 1% above 70%.[4]
When is wrist heart rate least accurate?
During irregular movement and work driven by the arms. No device in Gillinov 2017 met the accuracy threshold on an elliptical with moving arm levers, and only three were acceptable on a bike.[2] In Gielen 2026, stop-start walking increased errors for several devices, while hot and cold rooms did not.[3] The device matters too: the worst wearables in that study averaged 9-14 bpm off even at walking pace.[3]
Does an armband close the gap?
Mostly. Moving the same Polar optical sensor from the wrist to the upper arm cut the mean absolute error from 6.41 to 1.43 bpm against a Polar H10.[5] The upper arm moves less and has steadier blood flow, so an armband is the best optical option for intervals, though a chest strap remains the reference.
What does this mean for how you train?
- Easy runs, steady cycling, zone 2 work: a good wrist sensor is accurate enough on average; check that the trend matches how the session felt.[1]
- Intervals, circuits, rowing and elliptical work: use a chest strap or armband; this is where wrist readings lag and arm movement corrupts the signal.[2][3]
- HRV and recovery data: a chest strap, because beat-to-beat precision is the weak point of optical sensors.[6]
Make the reading trainable
A heart-rate number only means something against your zones. The Zone 2 Heart Rate Calculator, using the Karvonen heart-rate-reserve method for a 42-year-old with a resting heart rate of 48 bpm, returns a Zone 2 aerobic-base band of 126–139 bpm against an estimated maximum of 178 bpm — these numbers are computed live by the hub engine. The Heart Rate Zone Calculator builds the full five-zone spread for intervals and tempo. This article is part of the 2026 Wearable & AI-Coaching Accuracy vs Value Index; for a specific device head-to-head see the chest strap vs optical heart rate comparison.
Frequently asked questions
How accurate is wrist heart rate compared with a chest strap?
Close on average, unreliable for single readings. In controlled lab tests good watches err by about 1-5 bpm on average against ECG,[1][4] but individual readings can sit 15 bpm or more away from the strap,[4][7] and some devices averaged 9-14 bpm off even at walking pace.[3]
Is wrist heart rate accurate for running?
Usually, on steady runs. Treadmill running was where wrist sensors performed best in a 50-person ECG comparison (concordance 0.88-0.93 for most devices),[2] and an Apple Watch Series 6 and Garmin Forerunner 945 stayed within about 1-2 bpm of ECG during a maximal treadmill test.[1] Intervals and stop-start efforts are where wrist readings lag.[3]
When is wrist heart rate least accurate?
During irregular movement and arm-driven work. No optical device tested met the accuracy threshold on an elliptical with moving arm levers,[2] and intermittent walking increased errors for several devices compared with steady walking.[3]
Does an armband close the gap?
Mostly. The same Polar optical sensor erred by 1.43 bpm on the upper arm and 6.41 bpm on the wrist against a Polar H10 reference.[5] An armband is the best optical option, but a strap is still the reference.
Which heart-rate sensor should I use for HRV?
A single-lead ECG chest strap such as the Polar H10 ($104.95). HRV needs precise beat-to-beat timing, which is what optical sensors are weakest at.[6]
References
- 1 Accuracy of wrist-worn wearable devices for determining exercise intensity (Ho, Yang, Li) — Apple Watch Series 6 and Garmin Forerunner 945 vs ECG during maximal exercise testing — Digital Health, 8:20552076221124393 (2022)
- 2 Variable Accuracy of Wearable Heart Rate Monitors during Aerobic Exercise (Gillinov et al.) — Medicine & Science in Sports & Exercise, 49(8):1697-1703 (Polar H7 strap, Apple Watch, Garmin Forerunner 235, Fitbit Blaze, TomTom Spark and Scosche Rhythm+ vs ECG) (2017)
- 3 Accuracy of Optical Heart Rate Measurements for 10 Commercial Wearables in Different Climate Conditions and Activities: Instrument Validation Study (Gielen, Van Oost, Debard, et al.) — JMIR Formative Research (2026)
- 4 Heart Rate Estimation Using the Galaxy Watch During Maximal Cardiopulmonary Exercise Testing: Cross-Sectional Validation Study (Inoue et al.) — Galaxy Watch 6 vs Polar H10 — JMIR Cardio, 10:e81917 (2026)
- 5 Wrist-Worn and Arm-Worn Wearables for Monitoring Heart Rate During Sedentary and Light-to-Vigorous Physical Activities: Device Validation Study (Schweizer, Gilgen-Ammann) — Polar Vantage V2 wrist and Verity Sense arm vs Polar H10 — JMIR Cardio, 9:e67110 (2025)
- 6 Polar H10 Heart Rate Sensor — official specifications and price ($104.95, single-lead ECG) — Polar (2026)
- 7 Wrist-worn optical and chest strap heart rate comparison in a heterogeneous sample of healthy individuals and in coronary artery disease patients (Sartor, Gelissen, van Dinther, et al.) — BMC Sports Science, Medicine and Rehabilitation, 10:10 (2018)