The most accurate heart rate monitors, by what has been measured

Two physical principles, not seven brands

Before the ranking, the distinction that decides almost everything here.

Electrical sensors — chest straps — detect the electrical activity of the heart muscle directly. It is the same signal an ECG reads. There is no inference step.

Optical sensors — wrists, armbands, rings — shine light into the skin and infer pulse from how much comes back off moving blood. Skin tone, tattoos, ambient temperature, how tight the band is and how much the limb is moving all affect the measurement, because all of them affect the light.

One of those is a measurement and the other is an inference. That is why electrical straps win, and why researchers use them as the yardstick rather than as a subject.

The study that should decide your purchase

The 2025 Sensors anatomical-placement study (n = 28, 14 male and 14 female) did something unusually clever: it put three identical Whoop 4.0 unitson the same participants' wrist, forearm and upper arm simultaneously, alongside a Polar Verity Sense on the forearm and a Garmin Forerunner 55 on the wrist, all measured against a Polar H10.

Running the same sensor at three placements at once removes the brand variable entirely. Whatever differences appear are down to where the sensor sits, nothing else.

What it found, by intensity

At rest (5 minutes). Everything worked. Concordance above 0.99 for all devices, bias under 1 bpm, and the Verity Sense lowest at roughly 1.6 ± 0.8% MAPE. If you only ever look at resting heart rate, your existing watch is fine and you should close this page and keep your money.

Cycling warm-up (5 minutes). Separation begins. Whoop on the upper arm had the lowest error; the Verity Sense came in around 3.87 ± 1.92% MAPE. The wrist-worn Whoop and the Garmin showed limits of agreement spanning roughly ±10-19 bpm.

Burpees. Everything optical failed. Every device dropped below 0.50 concordance, and the wrist-worn Garmin showed a bias of approximately 33 bpm. Thirty-three beats is not a rounding error; it is the difference between an easy effort and a maximal one. If your training is explosive and stop-start, no optical device currently measured will serve you and you need a strap.

Modified Bruce treadmill protocol.The most running-relevant test, and the Verity Sense's best result: bias under 0.1 bpm, concordance 0.997. Whoop on the upper arm reached 0.992. Wrist devices sat at 0.85 or below.

The conclusion the authors drew

Proximal placements — upper arm and forearm — consistently outperformed the wrist, at every intensity. Not sometimes, not for one brand: consistently.

Which means the sentence most people need to read is this one: the upgrade is not a better watch, it is moving the sensor up your arm, and it costs about a hundred dollars and works with the watch you already own.

Why the H10 ranks first when the study cannot measure it

There is a genuine subtlety here worth naming. The H10 was the reference. So the study cannot tell you the H10's own error — everything else was measured against it, which mathematically fixes its error at zero within the study.

Being repeatedly chosen as the criterion by independent researchers is a strong signal, and we score it a 10 for that. But it is a different kind of evidence from a measured error bar, and we would rather explain the distinction than quietly let a 10 imply something it does not.

What we could not verify

Most of this roundup, honestly. Only the H10 and the Verity Sense appear in a peer-reviewed study. The Garmin HRM 600 and HRM 200, the Wahoo TRACKR and TICKR FIT, and the COOSPO H808S have no published validation we could find. They are all electrical straps or armbands using the same physical principles as the validated devices, which is a reasonable basis for expecting them to work — but expectation is not measurement, and we score them accordingly rather than assuming.

How to choose

If accuracy is the only thing that matters

The Polar H10. It is what the researchers use, it does the one job, and it costs a fraction of any watch here.

If you will not wear a chest strap

The Polar Verity Sense, and this is not a consolation prize — it is the device that measured a bias under 0.1 bpm on the treadmill protocol. Be honest with yourself about this: an unworn ECG strap measures nothing at all, and the best sensor is the one you actually put on.

If you do explosive or stop-start training

A chest strap, and only a chest strap. Every optical device measured fell below 0.50 concordance during burpees. There is no optical product that solves this, at any price.

If you want the cheapest thing that works on the right principle

The COOSPO H808S is an electrical strap with dual connectivity at roughly a quarter of the H10's price. The principle is right and the price is genuinely low. There is also no evidence behind it whatsoever, and we are not going to imply otherwise because it is cheap.

Connectivity, which is where people get caught out

ANT+ broadcasts to unlimited devices at once and is what most bike computers and gym equipment use. Bluetooth generally pairs one device at a time and is what phones and watches use. A strap with both pairs to your watch and your bike computer simultaneously; a Bluetooth-only strap makes you choose. The compatibility guide has the platform-by-platform detail.