How Accurate Are Fitness Trackers for Heart Rate During Exercise?

Fitness trackers are reasonably accurate at rest, but accuracy falls measurably during exercise, especially high-intensity workouts. At rest, mean errors are typically under 5 BPM. During vigorous running or interval training, errors commonly reach 10-20 BPM. For most recreational fitness goals, this is acceptable. For serious training or medical monitoring, it is not.

Why Exercise Degrades Fitness Tracker Accuracy

All major fitness trackers use optical PPG sensors that shine light into your wrist skin and detect changes caused by blood pulsing through capillaries. This works reliably when your wrist is still. The problem is exercise.

When you run or lift weights, your wrist moves. That movement creates vibration and changes in how the sensor contacts skin, generating noise in the optical signal. This noise is called a motion artifact, and it is the dominant source of error in exercise heart rate measurements.

The particular challenge is frequency overlap. During running, wrist movement occurs at roughly 2-4 Hz. A typical exercise heart rate of 120-180 BPM is 2-3 Hz. When motion artifact frequency matches heart rate frequency, the algorithm can confuse them, leading to readings that are wrong by exactly one "step" or that lock onto the cadence frequency instead of the cardiac frequency.

What the Research Shows: Device-by-Device

Apple Watch: Consistently the best performer among wrist-based devices in independent studies. A 2026 meta-analysis in npj Digital Medicine found Apple Watch showed stronger agreement with ECG criterion measures during exercise than competing brands. Mean errors during moderate running typically range from 4-8 BPM.

Garmin: Performs well for endurance sports. Garmin's Elevate and ELEVATE v5 sensors incorporate accelerometer-based motion artifact removal tuned specifically for running and cycling cadence. Mean errors of 5-10 BPM during moderate exercise are typical in validation studies.

Fitbit: Higher variability than Apple and Garmin during vigorous exercise. Studies have found Fitbit mean errors of 10-18 BPM during high-intensity running. Fitbit performs better at moderate intensities.

Samsung Galaxy Watch: Performance is closer to Garmin and Apple Watch in newer generations (Watch 4 and newer with the BioActive Sensor). Older Samsung devices showed higher exercise error rates.

Oura Ring: The Oura Ring is primarily designed for resting and sleep monitoring. Its exercise heart rate accuracy is generally lower than dedicated fitness trackers because the ring form factor limits the ability to do motion compensation algorithms tuned for vigorous exercise.

Polar chest strap: Mean errors under 2 BPM at all intensities. This is the clinical-grade reference standard for wearable exercise heart rate monitoring.

At What Exercise Intensity Does Accuracy Break Down?

The breakdown follows a pattern across almost all wrist-based devices:

Low intensity (walking, gentle cycling, yoga): Minimal degradation. Most trackers perform close to their resting accuracy (2-5 BPM mean error). The wrist is relatively stable and motion frequency is low.

Moderate intensity (jogging, brisk cycling, resistance training with steady pace): Meaningful degradation begins. Expect 5-10 BPM mean errors. Individual readings can spike higher.

High intensity (interval training, sprints, HIIT circuits, race pace running): Significant degradation. Mean errors of 10-20 BPM are common. The watch may lock onto stride cadence rather than heart rate, giving readings that are suspiciously round numbers like 160 or 180 BPM that match running cadence.

Cycling (steady power output): Better than running at equivalent intensity because wrist motion is less. Errors closer to 5-8 BPM even at high power outputs.

Swimming: Worst performance. Water ingress and the nature of arm motion during swimming creates severe motion artifact. Most fitness trackers explicitly note reduced accuracy during swimming.

The Cadence Lock Problem

One of the most common failure modes during running is cadence lock. The watch algorithm "locks on" to the frequency of your arm swing instead of your heartbeat. Because typical running cadence (160-180 steps per minute) overlaps with typical running heart rates (150-180 BPM), the two signals are indistinguishable if the motion artifact is strong enough.

When cadence lock occurs, your heart rate reading will appear stable and may even look plausible, which makes it especially misleading. The reading stays fixed at your cadence while your actual heart rate might be fluctuating significantly.

Some devices use additional sensors (gyroscopes, barometers) or different LED wavelengths to break this ambiguity, with varying success. For a technical explanation of how algorithms try to solve this, see PPG motion artifact detection.

Practical Accuracy Data: What Studies Measure

Independent studies typically measure accuracy using metrics like:

Mean Absolute Error (MAE): Average absolute difference from ECG reference. Lower is better.

Mean Absolute Percentage Error (MAPE): MAE expressed as a percentage of the reference value. A MAPE of 5% at a heart rate of 150 BPM equals a 7.5 BPM error.

Limits of Agreement (LoA): Using Bland-Altman analysis, the range within which 95% of individual readings fall relative to the ECG reference. Wide LoA means the device is unpredictable even if the average is close.

A landmark 2017 Stanford study (Shcherbina et al.) tested seven consumer wearables including Apple Watch, Fitbit, Garmin, and Basis Peak. They found mean absolute percentage errors ranging from 2.4% to 20.8% during exercise, with all devices performing substantially worse during vigorous activity.

More recent studies continue to show the same pattern: error at rest is small, error at high intensity is significant.

How to Get Better Accuracy From Your Fitness Tracker

You cannot eliminate exercise-related accuracy loss, but you can minimize it:

Wear it tighter during workouts: The most common advice from manufacturers and researchers alike. A snug fit reduces sensor movement. Not so tight as to restrict circulation, but noticeably tighter than you might wear it day-to-day.

Position above the wrist bone: One finger-width up from the wrist bone reduces the bobbing motion that is worst right at the joint.

Use workout-specific modes: Selecting the correct activity type tells the device to use algorithms optimized for that movement pattern.

Use optical HR as a guide, not gospel: If you are training by zones, treat the wrist reading as a general indicator rather than a precise number. A 5 BPM error rarely changes your training decision.

Pair a chest strap for important sessions: Most fitness trackers can receive Bluetooth heart rate data from third-party sensors and will display the chest strap data instead of the optical reading. For interval sessions or race simulations, pairing a Polar H10 or Garmin HRM-Pro gives you accurate data on your familiar watch display.

Should You Worry About the Inaccuracy?

For most fitness tracking goals, probably not. The errors involved rarely matter for:

• Tracking workout trends over time (the error is relatively consistent)
• Basic zone training where zone 2 vs. zone 3 separation is what matters
• Step counting and general activity monitoring
• Daily resting heart rate tracking

Where accuracy matters more:

• Competitive athletes training with precision
• Cardiac rehabilitation patients where specific heart rate targets are prescribed
• Anyone using heart rate to manage a medical condition

For clinical-grade PPG accuracy versus consumer device differences, the gap is significant and the clinical devices use different validation standards entirely.

The Future: Will Exercise Accuracy Improve?

The industry is actively working on better exercise accuracy through:

• New LED wavelengths (red and infrared penetrate deeper and are less affected by surface motion)
• More sophisticated sensor arrays with additional accelerometers and gyroscopes
• Machine learning algorithms trained on larger datasets of exercise heart rate signals
• Form factor innovations (earbuds, chest patches, arm bands) that have better contact stability than a wrist watch

Progress is real but incremental. The fundamental physics of optical sensing on a moving limb impose limits that better algorithms can push back but not eliminate. See wrist PPG accuracy limitations for a detailed look at the physics involved.

References

1. Shcherbina A, et al. "Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort." Journal of Personalized Medicine 7(2):3 (2017). doi:10.3390/jpm7020003

2. Gillinov S, et al. "Variable accuracy of wearable heart rate monitors during aerobic exercise." Medicine & Science in Sports & Exercise 49(8):1697-1703 (2017). doi:10.1249/MSS.0000000000001284

3. Wang R, et al. "Accuracy of wrist-worn heart rate monitors." JAMA Internal Medicine 177(1):100-101 (2017). doi:10.1001/jamainternmed.2016.7152

4. Robergs RA, et al. "Validity and reliability of wristwatch heart rate monitors in research and clinical populations." Journal of Sports Medicine and Physical Fitness 60(2):217-224 (2020). doi:10.23736/S0022-4707.19.09957-0

5. Muhtar MT, et al. "Accuracy of optical heart rate sensors during various forms of exercise." BMC Sports Science, Medicine and Rehabilitation 14:215 (2022). doi:10.1186/s13102-022-00613-1