Where to Wear Your Fitness Tracker: Tattoos, Hairy Arms and Placement

Last Updated: September 8, 2026

Where you wear your fitness tracker changes its heart rate error more than which brand you bought. In the one study that tested position directly, a Fitbit Inspire 2 worn three finger widths above the wrist bone had a movement error of 7.3% against a chest strap, versus 20.5% sitting one finger width up where most people wear it. Tattoos matter too, but not the way people think: over ink the sensor tends to return nothing rather than a slightly wrong number. Hairy arms, in the only study that graded them, made no difference. Cold hands, strap tightness and which wrist you use each have a measured answer. Here is what the peer-reviewed data actually shows, and what each brand's own support pages say.

Want the whole list with a verdict on each? Try the wearable accuracy factors tool, then read on for the story behind the numbers.

I started pulling this together after a post about moving a Fitbit three finger widths up the arm did far better on Reddit than anything else I had shared, and the first question under it was about hairy arms. That question was not in the research at all until I went looking for it. So this post is the placement finding, plus every "does this mess it up?" question people actually ask, checked against studies and against what Apple, Google, Garmin, Oura, WHOOP, Samsung and Polar tell their own customers.

Where to wear your fitness tracker: the fix with the biggest number

Vermunicht and colleagues (European Heart Journal Digital Health, 2025) put a Fitbit Inspire 2 on ten healthy adults, first one finger width above the wrist joint and then three, and compared it against a Polar H10 chest strap during rest and movement.

Position on the forearm	Error at rest	Error during movement	Agreement with the chest strap
One finger width above the wrist bone	11.5%	20.5%	0.59
Three finger widths above the wrist bone	3.4%	7.3%	0.92
Chest strap	Reference	Reference	1.00

The mechanism is boring, which is a good sign. One finger width up puts the sensor over the mobile bones and tendons of the wrist joint, which flex and rotate with every hand movement and lift the sensor off the skin. Three finger widths up sits on flatter soft tissue that barely moves when your hand does.

Ten people is a small study and nobody has replicated it yet, so treat it as a strong lead rather than a law. What makes it credible is that every manufacturer already tells you the same thing. Google's heart rate page says to wear a Fitbit (Google Health) device "a finger's width above your wrist bone" at rest and higher during exercise. Apple says "above the wrist bone (towards your elbow, not your hand)." Polar says "at least a finger's width." WHOOP says about an inch. Garmin and Samsung both say above the wrist bone.

One inconsistency worth knowing: Google's sleep-stages and product-care pages say two to three finger widths, while its heart rate page says one. Higher is the safer reading of that.

Does this affect accuracy? Twelve things people ask about

Each of these came up either in the Reddit thread or in the studies themselves. The verdicts are only as strong as the evidence behind them, and I have tried to say exactly how strong that is.

Question	Verdict	The number	Where it comes from
Hairy arms	Does not matter, in the one study	p = 0.29	Vermunicht 2025, n=30 cardiac rehab patients
Tattoos	Matters, as dropout	36% had total dropout at rest	Navalta and Bunn 2025, n=25
Skin tone	Average error unchanged; spread wider	Pooled bias null in all strata	Bent 2020, Mulholland 2025, meta-analysis 2024
Cold hands	Matters	Signal amplitude down 41%	Google-funded cooling study, n=21
Which wrist	No for heart rate and sleep, yes for steps	+1,253 steps a day on the dominant wrist	Park 2019, n=12
Too tight or too loose	Matters	23 to 47% better at the right pressure	Scardulla 2020, custom rig, n=17
Two devices at once	Does not matter	0 missing values	n=16, four devices at once
Sharing a bed	Untested	+21% limb movements	12 couples, sleep lab
Small or large wrists	Untested on current hardware	No data	One 2019 null
Ring finger and fit	Rotation matters, finger choice untested	Signal to -7.86 dB at 30 degrees	Ring rotation study, n=10
An old device	Never studied	0 of 249 validation studies	Umbrella review
Sweat, lotion, sunscreen	Brand guidance only	0 studies on lotion or sunscreen	Garmin, Polar, Samsung, Oura pages

Hairy arms. The cardiac rehabilitation arm of the Vermunicht study graded forearm hair on a four-point photographic scale for 30 patients and compared the accurate group with the inaccurate one. Hair density did not differ between them (p = 0.29). Skin type (p = 0.54) and BMI (p = 0.18) were null too. The study then tried to rescue the ten least accurate patients by cleaning the sensor, shaving part of the forearm and taping the watch in place, all at once, and that bundle helped three of the ten. So shaving has never been tested on its own. The brands split on this one: Samsung lists "body hair, dirt, or other objects" as things that block the light, WHOOP's official line is that hairy arms are fine, and Apple, Garmin, Google, Oura and Polar do not mention hair at all.

Tattoos. Navalta and Bunn (Sensors, 2025) put an optical sensor over a tattoo and over clear skin on the same arm in 25 people. At rest the error was 22.9% over ink versus 2.9% on clear skin, and nine of the 25 had the tattooed sensor drop to zero for stretches of time. Remove those dropouts and the tattooed reading was nearly normal. The failure is a sensor that stops reporting, not one that drifts. The counterintuitive part is that ink darkness and tattoo age did not predict who failed. Apple, Garmin and Polar all warn about tattoos; Samsung says to wear the Galaxy Ring on a finger without one. The fix is to move the sensor to clear skin.

Skin tone. This is the topic with the most heat and the least agreement, so here is the honest summary. The two best-powered studies, one with balanced skin tones across six devices and one using an objective colour meter, found no difference in average heart rate error for mainstream watches. A 2024 meta-analysis of 140,771 paired readings found the average bias was null in every skin tone group but the spread of errors was 2.2 times wider in dark skin. And in the colour-meter study, dark-skin participants were 36% of the sample yet supplied 50% of the unexplained missing data on an Apple Watch. The pattern is the same as tattoos: the effect lives in gaps, not in wrong averages.

Cold hands. Cooling the forearm with ice cut the optical pulse signal by 41% while leaving the timing of the beats intact, which is why cold devices drop readings rather than invent them. Warming a wrist for 15 minutes took a low-perfusion blood oxygen error from 4.1 points to zero, and skin tone did not change that. Apple, Google, Samsung, Oura and Polar all warn about cold; Garmin's advice is to warm up for five to ten minutes and take a reading before a workout. Winter runs and cold bedrooms are where you should expect gaps.

Which wrist. Both wrists worn at once: heart rate differed by 0.37 bpm, and sleep metrics were null across 65 nights. Steps are a different story. The dominant wrist logged 1,253 more steps a day in one study, and telling the app the wrong wrist moved activity totals by 22 to 26% in another, where 15.6% of supervised participants got the setting wrong. Pick one wrist and set it correctly.

Too tight or too loose. In a custom wrist rig with a pressure sensor, a loose fit failed at every intensity, and tuning the pressure to each person beat a one-size setting by 23 to 47%. A second study showed that too much pressure flattens part of the pulse wave. No one has tested a real consumer strap at graded notch settings, so the practical rule is the manufacturers': Apple says tighten for workouts and loosen afterwards, Google says snug but not constricting, and the Fitbit Air guidance is that a pinky finger should slide under the band.

Kygo sits on top of whatever your wearable reports, so when a reading is off because of position or cold, the pattern it draws with your food and sleep is off too. If you want to see which of your own readings are moving with your meals once the inputs are clean, the app is free to start on iOS and Android.

Two devices at once. Sixteen people wore two armbands and two watches simultaneously against a chest strap. No optical or mechanical interference, no missing values. Wear both if you want to compare.

Sharing a bed. In a sleep lab, sleeping next to a partner raised limb movements by 21% without changing total sleep or efficiency. Nobody has measured what that does to a wrist tracker, and no validation study reports whether participants slept alone. The predicted direction is more minutes scored awake, but that is a prediction.

Ring finger and fit. A ring rotated 30 degrees from its ideal position lost most of its signal, and doubling the LED power did not bring it back. Left versus right hand barely matters (94.8% agreement between hands). Which finger is best has never been tested, so Oura's advice to use the index finger with the sensor bumps on the palm side is manufacturer guidance, though it is consistent with the rotation finding: a ring that cannot spin is a ring that keeps its signal.

Small wrists, old devices, lotion. These three share a verdict: nobody has checked. Wrist circumference has one 2019 null and is listed as an uncontrolled variable in 2026 studies. Device age was not analysed in any of 249 validation studies in the largest umbrella review. Lotion and sunscreen have zero peer-reviewed tests at a wearable site, so Garmin's "avoid wearing sunscreen, lotion, and insect repellent under the watch" is an instruction, not a finding. It is still a sensible one.

Things that help, ranked by the size of the effect

What to do	The number	How strong the evidence is
Wear it three finger widths up the forearm	20.5% to 7.3%	One study, n=10; every brand agrees
Snug for workouts, loosen after	Up to 47% better	Custom rigs; Apple, Google and Polar agree
Armband or chest strap for arm-heavy sport	Error range 4x tighter	Replicated across two brands
Warm up before trusting a cold reading	Blood oxygen error 4.1 points to 0	Strong for the mechanism
Ring sensors on the palm side, snug	30 degrees of rotation is enough to lose it	One study; Oura and Samsung agree
Pocket the phone when pushing a cart or stroller	Wrist loses about 1 in 5 steps	Replicated for direction
One wrist, set correctly in the app	22 to 26% of activity	Strong
Charge in the shower, not in bed	47% of participants missing data by night five	Strong for the pattern

The armband result deserves a line of its own because it is the cleanest experiment in the set. Moghaddam and colleagues (Sensors, 2026) put three identical WHOOP 4.0 units on the wrist, forearm and upper arm of the same 28 people and ran them against a chest strap. Same sensor, same firmware, same person: on a graded treadmill the wrist agreement interval was about four times wider than the upper arm's. A Polar study replicated the pattern with a different brand. If you row, use an elliptical with arm levers, lift or do intervals, the wrist is the wrong place for the sensor, and I have written up the full activity-by-activity picture in heart rate accuracy by activity type.

The cart and stroller finding matters most for step counts. Wrist devices count steps from arm swing, so a hand fixed on a handle produces a 19.8% undercount in the peer-reviewed test and far worse in a conference abstract. A phone in your pocket lost only 6.4%. Walking speed and terrain do similar damage, which is covered in the step count accuracy factors post.

The charging one is less obvious. In a five-night study of 299 people wearing a Garmin, 22% were missing data on night one and 47% by night five, with battery the authors' best explanation. If your tracker charges on the nightstand at bedtime, it is measuring nothing on exactly the nights you most want.

Things that quietly hurt

Some conditions do not make the reading wrong so much as take it outside what the sensor or the model was built for.

Rowing, elliptical with arm levers and swimming beat every wrist sensor tested. Rowing ran 13.4% error at the wrist against 3.8% for walking on the same device, and a Garmin that read within 1 bpm on dry land read 44 bpm low in the pool.

Calories from a weights session are a model failure, not a sensor failure. In a 62-person study against a metabolic cart, heart rate tracked the ECG at r 0.96 to 0.97 during resistance training while energy expenditure read 116% high. Halve the number your watch gives you after lifting and you will be closer.

Slow walking and stairs break step counts. Below about 4 km/h step error jumps from roughly 7% to 40% across 21 devices, and neither Fitbit tested on stairs met the 10% threshold at any pace.

Your first nights with a new device are worse sleep than usual. Sleep onset went from 20 to 14 minutes and total sleep rose 12 minutes between night one and night two in a 45-person study, and the effect was just as strong at home as in a lab. Judge a new tracker after a week, and if you are trying to pick one, the sleep tracker accuracy comparison covers how the brands compare against a sleep lab.

Battery saver and low power modes turn measurement off rather than degrading it. Apple documents that Low Power Mode stops background heart rate and blood oxygen measurements entirely. The gap in your data looks identical to not wearing the watch.

What the brands say, side by side

This table is the manufacturer guidance as written on their current support pages, checked on September 7, 2026. It is what the companies instruct, not a measured effect.

Brand	Position	Tightness	Tattoos	Cold	Hair
Apple	Above the wrist bone, towards the elbow	Snug; tighten for workouts, loosen after	Can affect the sensor	Perfusion may be too low	Not mentioned
Google (Fitbit, Pixel)	One finger width at rest, higher for exercise	Snug but not constricting	Not mentioned	Can hinder accuracy during exercise	Not mentioned
Garmin	Above the wrist bone	Snug, should not shift	Can block the light	Warm up 5 to 10 minutes	Not mentioned
Samsung	Above the wrist	Not too tight	Ring: avoid tattooed fingers	Keep warm	Listed as an obstruction
Polar	At least a finger width up	Tighten for training, loosen after	Avoid placing over them	Circulation may be too weak	Not mentioned
WHOOP	About an inch above the wrist bone	Snug but comfortable	Not mentioned	Not mentioned	"Hairy arms are marked safe"
Oura (ring)	Index finger, sensors on the palm side	Snug, not tight; size down if between	Not mentioned	Cold fingers cause gaps	Not mentioned

One caution on all of this. None of the 2025 or 2026 flagship devices, including the Apple Watch Series 11, Pixel Watch 5, Fitbit Air, WHOOP 5.0, Oura Ring 4 against a sleep lab, Galaxy Watch 9 and Galaxy Ring, has an independent peer-reviewed validation yet. The placement, pressure, cold and motion findings above are about optical and accelerometer physics and carry across generations. A specific error figure for a specific older model does not, which is why the heart rate accuracy tool dates every device figure it shows.

Common questions

What is the single biggest thing I can do to make my wearable more accurate?

Move it higher up the forearm and keep it snug during exercise. In the one study that tested position, three finger widths above the wrist bone cut movement error from 20.5% to 7.3%. It was ten people, so a strong lead rather than a law, but every manufacturer's guidance points the same way.

Do hairy arms affect a fitness tracker?

In the only study that graded arm hair, hair density made no difference to accuracy (p = 0.29). Shaving was never tested on its own. Samsung lists hair as an obstruction; WHOOP says hairy arms are fine.

Do tattoos stop a wearable reading your heart rate?

Often, yes. In the one study, 36% of people had the sensor return nothing over ink at rest, and error was 22.9% over the tattoo versus 2.9% on clear skin. Ink darkness and tattoo age did not predict it. Move the sensor to clear skin.

Does skin tone affect wearable heart rate accuracy?

Not the average error, in the best-powered studies. The spread of errors is wider and the effect shows up as missing readings rather than wrong ones.

Does cold weather make my wearable less accurate?

Cold roughly halves the optical signal and devices drop readings rather than guess. Warm up before trusting a reading and expect gaps on winter runs.

Does it matter which wrist I wear it on?

Not for heart rate or sleep. For steps, the dominant wrist logs about 1,250 more a day and the wrong app setting moves activity by about a quarter, so pick one wrist and set it correctly.

How tight should a fitness tracker be?

Snug enough that it does not slide during a workout, loose enough that a pinky finger fits under the band at rest. Loose fits failed at every intensity in lab testing and too much pressure flattens the pulse wave. Tighten one notch for exercise and back off for sleep.

Are smart rings more accurate than watches?

At night, slightly, for resting heart rate. During daytime movement a ring placed 9th of 10 devices in one climate-chamber study. Fit matters more than form factor: a ring that rotates loses most of its signal.

The bottom line

The well-evidenced factors are boring and physical: where the sensor sits, how tight it is, how warm the skin is, whether your arm is swinging. The exciting ones mostly changed nothing when tested, including hair, which wrist, and skin tone on average error. Tattoos and cold are real, but they show up as missing data rather than wrong numbers, which means the graph looks fine while a third of the night is absent.

Once the inputs are right, the interesting question is what your readings move with. Kygo correlates your heart rate, HRV and sleep with what you ate and when, so a clean signal turns into something you can act on. It is free to start on iOS and Android.

---

Key sources: Vermunicht et al. 2025 (European Heart Journal Digital Health) for forearm position, hair density and the shaving bundle, n=10 and n=30; Moghaddam et al. 2026 (Sensors) for wrist versus upper arm, n=28; Navalta and Bunn 2025 (Sensors) for tattoos, n=25, no external funding; Scardulla et al. 2020 (Sensors) for contact pressure, n=17; Bent et al. 2020 (npj Digital Medicine) and Mulholland et al. 2025 for skin tone; the ten-device climate chamber study (JMIR Formative Research 2026), KU Leuven funded; the Google-funded forearm cooling study (Scientific Reports 2026); Park et al. 2019 for dominant versus non-dominant steps; the Garmin five-night missingness study (JMIR mHealth 2026); and the manufacturer pages from Google, Apple, Oura, Samsung and Polar, all read on September 7, 2026.

---

A note on what is deliberately left out: the old Fitbit "sleep sensitivity" setting, which swung sleep totals by over two hours in older studies, only ever existed on movement-only trackers with no heart rate sensor and is not in the Google Health app, so it is not a tip for any device sold today. The direction of the tattoo bias is also left unstated because the source paper's sign convention conflicts with its own dropout data.

Disclaimer: Kygo is a personal data aggregation and insights platform designed for informational purposes only. The information provided by Kygo, including correlations, patterns, and trends identified in your data, does not constitute medical advice, diagnosis, or treatment. Always consult a licensed healthcare provider with any questions regarding medical conditions.

Have you moved your watch up your arm or switched to an armband and seen the readings change? I would like to know whether the three-finger-width result holds up outside a lab.