Chest strap, armband or watch: how accurate heart-rate sensors are in the gym

Heart-rate sensors use one of two methods. A chest strap detects the electrical signal of each heartbeat through electrodes on the skin. An optical sensor, found in watches and armbands, shines light into the skin and estimates heart rate from changes in blood volume. Validation studies give a consistent answer: chest straps agree most closely with a clinical ECG in almost every type of exercise, while optical sensors are often accurate during steady running or cycling and become less reliable with arm movement, gripping and rapid changes in intensity. For gym classes built on strength work and intervals, a chest strap is the safest choice; a well-fitted armband is a reasonable alternative for steadier sessions.

How the two sensor types measure

An electrocardiogram (ECG) records the electrical activity that triggers each heartbeat. A chest strap works on the same principle with two electrodes: it detects each beat and calculates heart rate from the time between beats, so arm movement has little effect on the signal.

Optical sensors use photoplethysmography (PPG): an LED lights the tissue under the sensor, and a photodiode measures how much light returns. Each pulse of blood changes that amount slightly, and an algorithm turns the pattern into a heart rate. The signal is weak, and anything that moves the sensor against the skin or squeezes the blood vessels underneath adds noise. A 2020 study in npj Digital Medicine names three main sources of error: skin characteristics, motion, and signal crossover, where the sensor locks onto the rhythm of a repeated movement, such as running cadence, and reports it as heart rate.

What validation studies show by exercise type

A 2017 study in Medicine & Science in Sports & Exercise compared a Polar H7 chest strap, a forearm optical band and four wrist watches with ECG in 50 healthy adults on a treadmill, a bike and an elliptical trainer. The chest strap agreed almost perfectly with ECG. Most optical devices performed well on the treadmill, fewer were acceptable on the bike, and none was accurate on the elliptical when the arms pushed the handles.

A 2018 study in the same journal tested eight devices in 50 adults during graded cycling and three sets of four resistance exercises at a 10-repetition-maximum load. Only the chest strap and a pair of optical sports headphones were valid in both modes; none of the wrist devices was. Across all devices, heart rate was increasingly underestimated as intensity rose.

The npj Digital Medicine study found that optical error during activity was on average about 30% higher than at rest, and that devices differed most in how they responded to changes in activity. Steady, prolonged efforts were measured reasonably well. The authors also observed delays of 0 to 43 seconds between the ECG and the optical readings, sometimes fixed and sometimes varying within a session.

Armbands sit on the forearm or upper arm, away from the flexing tendons of the wrist. A 2019 field study of the Polar OH1 armband followed 70 trained athletes for about 390 hours against a chest strap. Agreement was high in running and cycling and lower in sports with heavy arm involvement. The limits of agreement ranged from about 7 to 20 beats per minute across sports, and agreement was weakest in tennis, kayaking and fitness training. Placement alone does not guarantee accuracy, though: the forearm band in the 2017 study was among the least accurate devices.

The table summarises these studies; individual models vary. A 2020 systematic review of 158 publications found that heart-rate accuracy depended on manufacturer and device type, and noted that devices are redesigned so often that reviews quickly date.

The evidence on skin tone is unsettled. The npj Digital Medicine study found no significant difference in accuracy across skin tones, but its sample was small and its method has been criticised, and the OH1 field study reported some association with skin tone. Manufacturers themselves list tattoos, an overly tight fit and wrist flexion among the factors that disturb optical readings.

What sensor error means for training zones

Sensor error matters when it is large compared with the width of a zone. With zones set at 10% steps of maximum heart rate, each zone spans fewer than 20 beats for most adults.

Illustration, not a prescription: a stored maximum of 185 beats per minute.

• Each zone is 18.5 beats wide. Zone 4 (80–90%) runs from 148 to 166 beats per minute.
• A true heart rate of 160 is 86% of maximum, in zone 4.
• A reading within ±5 beats (155–165) stays in zone 4.
• A reading within ±20 beats (140–180) can show anything from zone 3 to zone 5.

An error of about five beats, in line with an armband during steady laboratory work, rarely changes the zone. An error of 20 beats, which optical sensors can produce during arm-heavy work, can move the reading by more than a full zone. Because wrist devices tend to underestimate at high intensity, an optical reading is more likely to show a zone below the true one than above it.

Timing adds a second problem. In a 30-second sprint, heart rate itself lags behind the effort, and an optical sensor can add a delay of its own, so the peak usually appears after the work ends.

Choosing sensors for a group class

Pulz8 accepts any chest strap, armband or watch that broadcasts heart rate over the standard Bluetooth heart-rate service. On the shared board, a sensor problem usually looks different from a physiological change: a tile that freezes during a kettlebell set, drops abruptly mid-interval, or settles on a suspiciously steady number during skipping points to the sensor before the athlete. Heart rate does not normally drop sharply while the work continues.

Chest straps need moist electrodes and a snug fit; until sweat improves contact, dry electrodes can give erratic readings. Armbands should sit firmly on the forearm or upper arm, on skin without tattoos. A watch remains the weakest option for grip-heavy work.

Key points

1. Use chest straps for strength, rowing and interval classes, where optical sensors are least reliable.
2. Accept armbands for steady cardio, fitted firmly on the forearm or upper arm.
3. Treat wrist readings during gripping and pushing as approximate, and expect underestimation at high intensity.
4. Check the sensor before the athlete when a tile jumps, freezes or flatlines during continuous work.
5. Judge short intervals by effort, because both the heart and optical algorithms lag behind sudden changes.

Sources

• Gillinov S et al. Variable accuracy of wearable heart rate monitors during aerobic exercise. Medicine & Science in Sports & Exercise, 2017.
• Boudreaux BD et al. Validity of wearable activity monitors during cycling and resistance exercise. Medicine & Science in Sports & Exercise, 2018.
• Bent B, Goldstein BA, Kibbe WA, Dunn JP. Investigating sources of inaccuracy in wearable optical heart rate sensors. npj Digital Medicine, 2020.
• Hermand E et al. Validation of a photoplethysmographic heart rate monitor: Polar OH1. International Journal of Sports Medicine, 2019.
• Fuller D et al. Reliability and validity of commercially available wearable devices for measuring steps, energy expenditure, and heart rate: systematic review. JMIR mHealth and uHealth, 2020.

Heart rate is a training signal, not medical advice. Consumer heart-rate sensors are not medical devices; anyone with a heart condition should follow their doctor's guidance on monitoring exercise.