On a steady walk, run or bike ride, a good watch usually lands close to a chest strap. The error grows when your arms work hard or your effort changes fast. In a Stanford test of 7 wrist devices, 6 had a median heart rate error under 5% on a stationary bike. In a Cleveland Clinic test, not one wrist device was accurate on an elliptical once people grabbed the moving arm handles. So trust the steady middle of your workout. Be careful with the jumps.
I’m Shirley. I’m not a doctor, a nurse or a pharmacist. A fitness watch is not a medical device for diagnosing anything. If you have a heart condition, your doctor is the one who should tell you which numbers matter for you.
I hear this question a lot. Usually it’s from someone whose watch showed a number that didn’t match how they felt (too high on an easy hill, or oddly low while they were huffing on a machine). So here’s how the watch gets its number, how far off it can be, and when it’s worth checking another way.
How the Light on the Back of Your Watch Finds Your Pulse
Turn your watch over. Those little lights on the back are the whole trick.
Apple’s support page explains it plainly. Its watch pairs green LED lights with light-sensitive sensors and flashes them hundreds of times per second. Each heartbeat pushes a little more blood through your wrist, and blood soaks up green light, so the watch counts those tiny changes as beats. Apple lists the sensor’s range as 30 to 210 beats per minute.
That method has a long name: photoplethysmography. Everyone shortens it to PPG, and most fitness watches and bands use some version of it.
A chest strap works differently. It picks up the electrical signal of your heart itself, a lot like the stickers on your chest during an ECG at your doctor’s office. That’s why researchers use a chest strap or an ECG as the yardstick when they test watches. Your watch is reading blood flow at the wrist, while the strap is reading the heart.
What the Studies Found at Rest and on the Move
Three studies are worth knowing about. They mostly agree.
In 2017, a Stanford team strapped 7 wrist devices on 60 volunteers, 29 men and 31 women, with an average age of 38. Everyone sat, walked, ran and cycled while a continuous heart monitor kept score. Six of the 7 devices had a median heart rate error under 5% on the bike. Walking had the highest error of any activity. The Apple Watch did best overall; the Samsung Gear S2 did worst.
Engineers at Duke came at it from another angle in 2020. They tested consumer watches next to research-grade wearables, at rest and during activity, and found that the error during activity was, on average, 30% higher than at rest. Their summary is a fair one. The devices were all reasonably accurate at rest and during a long, steady effort, but they differed in how they handled a change in activity.
Skin tone is where the two studies part ways. Stanford found more error with darker skin, and also in men, in people with a higher body mass index and during walking. Duke looked across the full range of skin tones and found no statistically significant difference. Honestly, I’d rather tell you that plainly than pick the answer I like. It isn’t settled.
READ: Calories Burned on Your Fitness Tracker: How Accurate Is the Number?
Where the Error Gets Big
If you use a gym machine, this is the study I’d read. In 2017 the Cleveland Clinic tested 50 healthy adults (average age 38, 54% women) on a treadmill, a stationary bike and an elliptical trainer, the elliptical both with its arm handles and without. Every reading was checked against an ECG at rest and at light, moderate and vigorous effort.
They gave each device an agreement score. A 1 would be a perfect match with the ECG.
The chest strap scored 0.996, which is about as close as it gets. Apple Watch came next at 0.92, then the TomTom Spark at 0.83 and the Garmin Forerunner at 0.81. A forearm band, the Scosche Rhythm+, managed 0.75. The Fitbit Blaze came last at 0.67.
The machine mattered just as much as the brand. On the treadmill, most devices scored between 0.88 and 0.93, and only the Fitbit Blaze fell behind, at 0.76. On the bike, only 3 of the wrist and forearm devices were good enough. On the elliptical with the arm handles moving? None of them. Every single one scored under 0.80.
Apple’s own fine print says much the same thing. Rhythmic movements like running or cycling give better results than irregular ones like tennis or boxing. Lifting weights wasn’t part of these studies, so I can’t give you a number for it. My guess is that a lot of stop-and-start gripping won’t help, but that’s a guess, not a reading.
Cold Hands, a Tattoo or a Loose Band
Some things that throw off the reading have nothing to do with the watch. Apple lists them on its support pages, and they apply to pretty much any watch that reads your pulse with light.
Start with blood flow in your skin. It varies a lot from one person to the next, and the weather changes it too. If you’re exercising in the cold, Apple says, the blood flow in your wrist might be too low for the sensor to get a reading at all. So cold hands on a January walk are a real reason for a strange number, not a sign that your heart is doing something odd.
Tattoos come next. Apple says the ink, the pattern and how saturated the color is can block light from the sensor.
Then there’s fit, and it’s the one you can fix today. Apple wants the watch snug on top of your wrist with the sensor close to your skin, and it even suggests tightening the band for workouts and loosening it afterward. A watch that slides around can’t keep the sensor against your skin. Turns out a single notch makes a difference.
Chest Strap or Watch: Which One You Need
For most people on a walk or an easy bike ride, the watch is fine. What matters is how much an error would matter to you.
The Cleveland Clinic team didn’t hedge. When an accurate heart rate is important, they wrote, use a chest monitor with electrodes. A second study from the same group in 2019 tested 80 patients in cardiac rehab, on a treadmill and a stationary bike. The chest strap scored 0.99 against the ECG, while the best watch managed 0.80 and the weakest just 0.52. None of the wrist devices was as accurate as the strap.
So if your doctor or your rehab team gave you a heart rate you shouldn’t go over, I’d wear a chest strap during exercise. Most watches can pair with one over Bluetooth. Apple suggests exactly that when the wrist can’t get a steady reading.
READ: Heart Rate Monitor: Which Should I Choose?
How to Get a Better Reading From Your Watch
You can’t turn a watch into a chest strap. You can get a lot closer to what the studies saw, though.
Tighten the band one notch before you start, and wear it on top of your wrist. On a cold day, give your hands a few minutes to warm up. Read the display during the steady part of your workout rather than in the first minute after you speed up, because a change in effort is when the Duke team saw the devices differ the most.
Two checks need no gadget at all. You can count your pulse at your wrist for 30 seconds and double it, the way the NHS explains it. Or use the CDC’s talk test: during moderate exercise you can talk but not sing, and during vigorous exercise you can’t say more than a few words without stopping for a breath. If the display and your breathing disagree, believe your breathing.
READ: Heart-Rate Zones & Maximum Heart Rate
When the Number Means Call Your Doctor
A watch can be wrong. Your body can’t.
If you feel chest pain or pressure, dizziness, unusual shortness of breath or a fluttering, uneven heartbeat while you exercise, stop. Don’t look at the watch to decide whether it’s real. Call 911 for chest pain that doesn’t go away, and call someone rather than drive yourself.
At rest, Mayo Clinic gives the normal range for adults as 60 to 100 beats per minute. If your watch keeps showing a resting number outside that, and especially if you also feel unwell, tell your doctor. Bring the readings along and mention which watch you wear. And if you take any heart medicine, ask your doctor which numbers to aim for during exercise. The usual charts may not fit you.
The Short Version
Your watch reads blood flow with light. On a steady walk, run or ride it’s usually close: 6 of 7 wrist devices stayed under a 5% median error on a bike, and the best one scored 0.92 against an ECG. The error runs about 30% higher when you’re moving than when you’re still, and it gets much worse with arm handles, irregular moves, cold hands, a tattoo or a loose band.
If an exact number matters for your health, wear a chest strap and ask your doctor what to aim for. If you feel unwell, stop. Whatever the display says.
