Can a Smartwatch Measure Your Blood Pressure? The FDA-Cleared Feature Never Shows You a Number

Your watch buzzes. Possible hypertension. You have not felt anything, you have no cuff in the house, and the notification does not tell you a number — not 130, not 145, nothing. Just a word.

Here is the honest answer to the question underneath that notification: no consumer smartwatch measures your blood pressure. The one feature the FDA has cleared for hypertension on a watch was designed, deliberately, never to show you a reading. It watches for a pattern over 30 days and tells you to go find a cuff.¹

And the asymmetry inside it runs the opposite direction from what most people assume.

An alert is worth acting on. The absence of one tells you far less than it feels like it does. In Apple’s own pivotal study, the feature flagged 41.2% of people who actually had hypertension.² Nearly six in ten went unflagged. Applied to the US population, a modelling study in JAMA found that among adults eligible for the feature, someone who receives no alert still carries a 21% chance of having undiagnosed hypertension — and if they’re 60 or older, 34%.³

This guide works from the primary record: the FDA clearance and its indications, the published sensitivity and specificity, the population modelling, and the validation literature on why cuffless blood pressure is such a hard problem. None of it is medical advice, and none of it replaces a clinician.

Can a smartwatch measure your blood pressure?

No. As of August 2026, no consumer smartwatch produces a blood pressure reading that a guideline body accepts for diagnosis or management. Two very different things ship on wrists today, and conflating them is the source of nearly all the confusion.

Approach	What it gives you	Regulatory status	What it explicitly won’t do
Apple Watch Hypertension Notifications	A notification only — never a systolic/diastolic number	FDA 510(k) K250507, cleared September 11, 2025. Over-the-counter, adults 22+, no prior hypertension diagnosis¹	Not for diagnosis, not for monitoring treatment effect, not for blood pressure surveillance¹
Samsung Galaxy Watch Blood Pressure	An actual systolic/diastolic number, after cuff calibration	Not FDA-cleared. Launched in the US March 31, 2026 under the FDA’s general wellness policy for low-risk devices⁴	Not intended for diagnosis, cure, mitigation, treatment, or prevention of disease⁴
Everything else (Oura, WHOOP, Garmin, Fitbit)	No blood pressure feature at all	No clearance, no claim	—

Read those two right-hand columns together and the picture resolves. Apple got a clearance precisely because it refused to claim a measurement. Samsung shows you a measurement precisely because it isn’t claiming to be a medical device. Nobody has done both.

That matters more than it sounds, because hypertension is the condition most worth catching early and easiest to miss. In NHANES data covering August 2021 through August 2023, 47.7% of US adults had hypertension, and only 59.2% of them were aware of it.⁵ Roughly four in ten people with high blood pressure don’t know. A wrist-worn screen aimed at that gap is a genuinely good idea. The question is what its numbers actually support.

What the Apple Watch hypertension notification actually does

It runs in the background and never asks you for anything. The feature analyses photoplethysmography data — the green optical sensor reading blood volume changes under your skin — collected opportunistically while you wear the watch during waking hours. It looks for patterns consistent with chronic high blood pressure across discrete 30-day windows, and speaks up only when the pattern persists.¹

Think of it less like a blood pressure cuff and more like a smoke alarm in a neighbouring room. It isn’t measuring the fire. It’s noticing that something in the building has been off for a month.

The indications for use are worth reading in the FDA’s own framing rather than a headline’s. The feature is intended for over-the-counter use by adults 22 and older who have not been previously diagnosed with hypertension, and it is explicitly not intended to replace traditional methods of diagnosis, to monitor the effect of hypertension treatment, or to serve as a method of blood pressure surveillance.¹

So if you already know you have high blood pressure, this feature is not built for you. If you’re on medication, it is not built for you either. It is a one-way screening tripwire for people who have never been told they have the condition.

What does an alert mean — and what does no alert mean?

This is the part where the numbers get genuinely interesting, and where most coverage stops one step too early.

Apple’s pivotal study enrolled more than 2,000 adults without a prior hypertension diagnosis. Participants wore the watch at least 12 hours a day for 30 days while taking twice-daily home readings with a validated cuff as the reference standard. The result: sensitivity 41.2% (95% CI, 37.2–45.3) and specificity 92.3% (95% CI, 90.6–93.7), with sensitivity rising to 53.7% (95% CI, 47.7–59.7) in people with stage 2 hypertension.²

That is a deliberate design trade. Screening tests usually maximise sensitivity so cases aren’t missed; this one maximises specificity so alerts aren’t wasted. It buys a trustworthy alert at the cost of a large silent group.

A research letter in JAMA by Jordana B. Cohen, MD, MSCE, a nephrologist and hypertension researcher at the University of Pennsylvania’s Perelman School of Medicine, working with Daichi Shimbo, MD of Columbia University Irving Medical Center and Adam P. Bress, PharmD, MS of the University of Utah, took those figures and applied them to a nationally representative NHANES sample. They identified 3,983 participants matching the feature’s eligibility criteria — 22 or older, not pregnant, unaware of any hypertension diagnosis, not on antihypertensive medication — representing 127 million US adults.³

Their finding: a positive predictive value of 69.1% (95% CI, 63.3–74.9) and a negative predictive value of 79.0% (95% CI, 76.6–81.3).³

Here is what that means in the only terms that matter to you — the probability you have hypertension before and after the watch speaks:

If you’re…	Chance you have undiagnosed hypertension	After an alert	After no alert
Any eligible adult	30%	69%	21%
Under 30	14%	47%	10%
30–44	26%	65%	18%
45–59	36%	75%	27%
60 or older	45%	81%	34%
BMI over 30	39%	78%	29%
BMI 30 or under	24%	63%	17%

Source: Cohen et al., JAMA 2026, modelled on NHANES 2017–2020.³

Two things fall out of that table.

An alert is a real signal. It roughly doubles your probability of having hypertension, from 30% to 69% overall. For a 60-year-old, an alert takes it to 81%. That is worth a cuff and a doctor’s visit, promptly.

No alert is not an all-clear. A quiet month still leaves a one-in-five chance overall, and a one-in-three chance if you’re over 60. Under 30, an alert is close to a coin flip — 47% — because the underlying condition is rarer at that age, and low prevalence punishes even a specific test.

Cohen and colleagues put the risk plainly: “A large proportion of individuals unaware of their hypertension may be made aware. However, an even larger proportion of individuals with undiagnosed hypertension could receive no alert at all,” and “false reassurance may discourage some individuals with undiagnosed hypertension from obtaining appropriate screening or engaging with the health care system.”³

That framing — a signal you should act on, a silence you shouldn’t trust — is the same asymmetry we walked through for smartwatch sleep apnea screening, except reversed. Apple’s apnea feature has high specificity and an explicit warning that absence means nothing. The hypertension feature has the same structural property, and it deserves the same reading. Turning a probability into a next step is a different job from generating one, and it’s the job SensAI is built for across the rest of your health data.

Why is cuffless blood pressure so hard?

Because a cuff measures pressure and a watch infers it.

An inflatable cuff physically occludes an artery and detects when blood starts moving again. It’s a direct mechanical measurement. A wrist sensor sees a pulse waveform and runs a model that maps waveform shape to a pressure number. That model has to be anchored somewhere — and the anchor is a cuff reading you supply during calibration.

The best-resourced attempt to make that work failed. Ramakrishna Mukkamala, PhD, of the University of Pittsburgh, together with George S. Stergiou, MD — chair of the European Society of Hypertension’s Working Group on Blood Pressure Monitoring — reviewed Microsoft Research’s Aurora Project, which they described as the most important and highest-resource assessment of cuffless devices to date. Across 1,125 participants testing several pulse-wave-analysis devices, “the overall results were clear-cut negative.”⁶

The failure mode is specific and it is instructive. When researchers led by Maarten Falter at Hasselt University put a Samsung Galaxy Watch Active 2 on 40 patients alongside 24-hour ambulatory monitoring, they found a systematic bias toward the calibration point: low blood pressures were overestimated and high ones underestimated.⁷ The watch pulls every reading back toward the number you fed it.

Put another way: the device tends to tell you what you told it. And it does so most strongly at exactly the values you’d most want it to catch.

A 2026 prospective validation offers the fairest version of the counter-argument. Stanislav Walzel and colleagues at the Czech Technical University in Prague ran a 28-day protocol with 37 participants, comparing a Galaxy Watch 5 against a validated reference sphygmomanometer every day across a full calibration cycle. The watch met ISO 81060-2 and IEEE 1708 accuracy criteria on all but one measure, with negligible mean differences (−0.34 mmHg systolic, 0.62 mmHg diastolic) and minimal drift over the month. But when the reference reading sat 10 mmHg away from the calibration point, the mean difference grew to 3.4 mmHg systolic and 5.1 mmHg diastolic, and the authors concluded that cuff confirmation is advised whenever blood pressure fluctuates substantially or a clinical decision is planned.⁸

That is the honest state of the technology. Near your calibration point, a modern cuffless watch tracks well. As your blood pressure moves away from that point — which is the entire event you’re monitoring for — accuracy degrades.

The European Society of Hypertension has been consistent about what follows. Its working group statement notes that ESH guidelines do not recommend cuffless devices for the diagnosis and management of hypertension, and that validation protocols built for cuff devices are inadequate for cuffless ones.⁹ Its 2023 follow-up laid out what a proper cuffless validation would have to test — the need for individual cuff calibration, stability after calibration, the ability to track blood pressure changes, and the behaviour of the underlying machine learning.¹⁰ A 2026 review in the European Journal of Preventive Cardiology led by Gianfranco Parati reached the same place on the technical foundations.¹¹

What about the number on a Galaxy Watch?

Samsung began a phased US rollout of blood pressure monitoring on March 31, 2026, for Galaxy Watch 4 and later. It requires the Samsung Health Monitor app and an initial calibration with a third-party cuff, with recalibration every 28 days.⁴

Critically, Samsung shipped it in the US as a general wellness feature, not a medical device — which is why it doesn’t carry FDA clearance. Samsung’s own framing is that the feature is not intended for use in diagnosing disease or in the cure, mitigation, treatment, or prevention of disease.⁴

So the number on your Galaxy Watch is real in the sense that a model produced it. It is not a clinical measurement, it requires you to already own the cuff it’s anchored to, and its accuracy is best where your blood pressure already was 28 days ago.

If you find that unsatisfying, that’s the correct reaction. It is the same category of caveat that applies to wearable calorie burn estimates — a modelled number presented with the confidence of a measured one.

How should you actually check your blood pressure?

With a validated upper-arm cuff, measured properly, more than once. The technique matters as much as the device — the American Heart Association’s scientific statement on blood pressure measurement, led by Paul Muntner, documents how much error comes from posture, cuff size, talking, and inadequate rest before the reading.¹²

The practical protocol most guidelines converge on:

Sit quietly for 5 minutes before measuring — back supported, feet flat, arm at heart level.
Use an upper-arm cuff sized to your arm. A cuff that’s too small reads high.
Take two readings a minute apart, morning and evening, and average them.
Do this for 7 days before drawing a conclusion. One reading is noise.
Bring the log to your clinician rather than a screenshot of a notification.

There is strong evidence that the out-of-office numbers are the ones that predict outcomes. In an observational cohort of 59,124 primary care patients followed for a median of 9.7 years, Neil Staplin and colleagues found 24-hour ambulatory systolic blood pressure was more strongly associated with all-cause death (HR 1.41 per 1-SD increment) than clinic systolic pressure (HR 1.18) — and after adjusting for ambulatory readings, the clinic association nearly vanished.¹³ Masked hypertension, where clinic readings look fine but out-of-office readings don’t, carried elevated mortality risk; white-coat hypertension did not.¹³

That’s the strongest argument for wrist-based screening in principle. The problem your watch is aiming at — blood pressure that misbehaves outside the clinic — is genuinely the one that kills people. The technology just isn’t there yet.

The 2025 AHA/ACC hypertension guideline remains built on validated cuff-based measurement for detection and management.¹⁴ Nothing on your wrist changes that in 2026.

What your watch is genuinely good at

None of this makes the device on your wrist useless. It makes it a different instrument than people think it is.

Wearables are strong at trends in things they measure directly: resting heart rate, heart rate variability, sleep duration and consistency, training load, and how quickly your heart rate falls after exercise. Those are direct optical and temporal measurements, not inferred pressures — which is why heart rate recovery and resting heart rate by age hold up far better as personal signals than any cuffless blood pressure estimate does.

They’re also strong at adherence, which is the variable that actually moves blood pressure. A watch that keeps you training four days a week is doing more for your cardiovascular risk than any wrist-derived systolic number would.

This is the split SensAI is built around: use the wearable data that’s actually validated, in context, and be explicit about what the hardware can’t see. Reading a single day’s numbers as a verdict is the most common mistake in wearable interpretation — the same trap we mapped in detail for HRV accuracy across devices. SensAI treats your recovery and load data as a trend against your own baseline, and it will never pretend to know a number your hardware didn’t measure.

What actually lowers blood pressure

If a notification sent you here, the useful question isn’t which watch to buy. It’s what moves the number once a cuff confirms it.

The largest synthesis to date is a network meta-analysis by Jamie J. Edwards and Jamie M. O’Driscoll at Canterbury Christ Church University, pooling 270 randomised controlled trials and 15,827 participants.¹⁵ Every mode worked. They did not work equally.

Training mode	Change in resting BP (systolic/diastolic)	SUCRA rank for systolic
Isometric exercise (e.g. wall squats)	−8.24 / −4.00 mmHg	98.3%
Combined aerobic + resistance	−6.04 / −2.54 mmHg	75.7%
Dynamic resistance training	−4.55 / −3.04 mmHg	46.1%
Aerobic exercise	−4.49 / −2.53 mmHg	40.5%
High-intensity interval training	−4.08 / −2.50 mmHg	39.4%

Source: Edwards et al., British Journal of Sports Medicine 2023.¹⁵

Isometric training — holding a static contraction rather than moving through reps — came out on top, with the isometric wall squat the single most effective submode for systolic pressure.¹⁵ It’s an unglamorous finding and an easy one to act on. We’ve broken down the full protocol in our guide to exercise to lower blood pressure.

Worth noting what the table doesn’t say: it doesn’t say to drop everything else. Every mode in it lowered blood pressure. If a clinician has confirmed high readings, tell SensAI — it will fold isometric holds into a program you’ll actually keep doing rather than handing you a protocol that competes with your training.

Frequently asked questions

Can an Apple Watch measure blood pressure?

No. The Apple Watch has no blood pressure measurement feature and does not display a systolic or diastolic reading. Its FDA-cleared Hypertension Notifications feature analyses 30 days of optical sensor data and sends a notification if it detects patterns suggestive of chronic hypertension — nothing more.¹

How accurate is the Apple Watch hypertension notification?

In Apple’s pivotal study of more than 2,000 adults, sensitivity was 41.2% (95% CI, 37.2–45.3) and specificity 92.3% (95% CI, 90.6–93.7). Sensitivity was higher — 53.7% — for stage 2 hypertension.² It rarely fires falsely, but it misses roughly six in ten cases.

What should I do if I get a hypertension notification?

Confirm it with a validated upper-arm cuff and see a clinician. A modelling study found an alert raises the probability of undiagnosed hypertension from about 30% to 69% in eligible adults, and to 81% in those 60 and older.³ It is a prompt to get measured, not a diagnosis.

Does no notification mean my blood pressure is fine?

No. Among eligible US adults, the absence of an alert still leaves about a 21% chance of undiagnosed hypertension, rising to 34% in adults 60 and older.³ Researchers have specifically warned that false reassurance may keep people from getting screened.³

Can a Samsung Galaxy Watch measure blood pressure?

It displays a number, but in the US it ships as a general wellness feature without FDA clearance, requires calibration against a separate cuff, and must be recalibrated every 28 days.⁴ Samsung states it is not intended for diagnosis or treatment of disease.⁴

Are cuffless blood pressure devices accurate?

Not reliably. Microsoft Research’s Aurora Project tested several pulse-wave-analysis devices in 1,125 participants and the results were “clear-cut negative.”⁶ The European Society of Hypertension does not recommend cuffless devices for diagnosing or managing hypertension.⁹

Why do smartwatch blood pressure readings drift toward my calibration value?

Because the device models pressure from a pulse waveform anchored to the cuff reading you supplied. One validation found systematic bias toward the calibration point — low pressures overestimated, high ones underestimated.⁷ A 2026 study found accuracy degraded as readings moved 10 mmHg away from the calibration point.⁸

Is a blood pressure watch worth buying?

Not as a substitute for a cuff. A validated upper-arm monitor costs less than most smartwatches and is what every guideline body accepts. If you already own a watch with a hypertension notification feature, treat it as a free background screen with a useful alert and an untrustworthy silence.

Which exercise lowers blood pressure most?

Isometric training. A network meta-analysis of 270 trials and 15,827 participants found isometric exercise produced the largest reductions (−8.24/−4.00 mmHg), ranking first among all modes, with the wall squat the most effective submode.¹⁵

How often should I check my blood pressure at home?

Take two readings a minute apart, morning and evening, for seven consecutive days, and average them — measurement technique drives a large share of error.¹² Out-of-office readings predict mortality better than clinic readings.¹³

The bottom line

Your watch cannot measure your blood pressure. What Apple built instead — a passive 30-day pattern detector that never shows a number — is a reasonable piece of engineering with a clear-eyed regulatory description of its own limits, and it will genuinely find people who didn’t know they had hypertension.

It will also stay silent for most of them. A 41.2% sensitivity means the feature’s quiet is not evidence of anything, and the group most harmed by treating it as evidence is people over 60, where a silent month still carries a one-in-three chance.

If you get an alert, get a cuff and get to a doctor. If you don’t get one, still get a cuff. And whatever the number says, the intervention that moves it is the same one your watch is actually good at tracking: showing up, consistently, to train.

References

U.S. Food and Drug Administration. “510(k) Premarket Notification: Hypertension Notification Feature (HTNF), Apple Inc.” K250507, cleared September 11, 2025. https://www.accessdata.fda.gov/cdrh_docs/pdf25/K250507.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶
Apple Inc. “Hypertension Notification Feature on Apple Watch.” Validation study, September 2025. https://www.apple.com/health/pdf/Hypertension_Notifications_Validation_Paper_September_2025.pdf ↩ ↩² ↩³
Cohen JB, Addo DK, Jacobs JA, Greene TH, Xu Y, Picone DS, Shimbo D, Bress AP. “Impact of a Smartwatch Hypertension Notification Feature for Population Screening.” JAMA, 2026; 335(11): 1001-1003. https://pubmed.ncbi.nlm.nih.gov/41661624/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸
Samsung Electronics. “Samsung’s Blood Pressure Monitoring Feature Now Available to U.S. Users in Samsung Health.” Samsung Newsroom U.S., March 31, 2026. https://news.samsung.com/us/samsung-blood-pressure-monitoring-feature-available/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶
Fryar CD, Kit B, Carroll MD, Afful J. “Hypertension Prevalence, Awareness, Treatment, and Control Among Adults Age 18 and Older: United States, August 2021–August 2023.” NCHS Data Brief, 2024; no. 511. https://pubmed.ncbi.nlm.nih.gov/40085792/ ↩
Mukkamala R, Shroff SG, Landry C, Kyriakoulis KG, Avolio AP, Stergiou GS. “The Microsoft Research Aurora Project: Important Findings on Cuffless Blood Pressure Measurement.” Hypertension, 2023; 80(3): 534-540. https://pubmed.ncbi.nlm.nih.gov/36458550/ ↩ ↩²
Falter M, Scherrenberg M, Driesen K, Pieters Z, Kaihara T, Xu L, Caiani EG, Castiglioni P, Faini A, Parati G, Dendale P. “Smartwatch-Based Blood Pressure Measurement Demonstrates Insufficient Accuracy.” Frontiers in Cardiovascular Medicine, 2022; 9: 958212. https://pubmed.ncbi.nlm.nih.gov/35898281/ ↩ ↩²
Walzel S, Sebestova H, Rafl-Huttova V, Rozanek M, Rafl J. “Long-term accuracy and stability of blood pressure measurements from a smartwatch: Prospective validation study.” Digital Health, 2026; 12: 20552076261415923. https://pubmed.ncbi.nlm.nih.gov/41602947/ ↩ ↩²
Stergiou GS, Mukkamala R, Avolio A, Kyriakoulis KG, Mieke S, Murray A, Parati G, Schutte AE, Sharman JE, Asmar R, McManus RJ, Asayama K, De La Sierra A, Head G, Kario K, Kollias A, Myers M, Niiranen T, Ohkubo T, Wang J, Wuerzner G, O’Brien E, Kreutz R, Palatini P. “Cuffless blood pressure measuring devices: review and statement by the European Society of Hypertension Working Group on Blood Pressure Monitoring and Cardiovascular Variability.” Journal of Hypertension, 2022; 40(8): 1449-1460. https://pubmed.ncbi.nlm.nih.gov/35708294/ ↩ ↩²
Stergiou GS, Avolio AP, Palatini P, Kyriakoulis KG, Schutte AE, Mieke S, Kollias A, Parati G, Asmar R, Pantazis N, Stamoulopoulos A, Asayama K, Castiglioni P, De La Sierra A, Hahn JO, Kario K, McManus RJ, Myers M, Ohkubo T, Shroff SG, Tan I, Wang J, Zhang Y, Kreutz R, O’Brien E, Mukkamala R. “European Society of Hypertension recommendations for the validation of cuffless blood pressure measuring devices.” Journal of Hypertension, 2023; 41(12): 2074-2087. https://pubmed.ncbi.nlm.nih.gov/37303198/ ↩
Parati G, Ochoa JE, Abreu A, Brouwers S, Bruno RM, Caiani EG, Casado-Arroyo R, Cluitmans M, Davos CH, De Lucia R, Dilaveris P, Guerra F, Hanssen H, Jensen MT, Kahan T, Kemps H, Leeson P, Locati ET, Lumens J, Mahfoud F, Roig JM, Omboni S, Papaioannou TG, Platonov PG, Ramirez J, Schuuring MJ, Sudano I, Treskes RW, Avolio A, Castiglioni P. “Cuffless Blood Pressure Monitoring Devices: Technical Foundations and Clinical Implications.” European Journal of Preventive Cardiology, 2026; 33(7): 1058-1071. https://pubmed.ncbi.nlm.nih.gov/41739846/ ↩
Muntner P, Shimbo D, Carey RM, Charleston JB, Gaillard T, Misra S, Myers MG, Ogedegbe G, Schwartz JE, Townsend RR, Urbina EM, Viera AJ, White WB, Wright JT Jr. “Measurement of Blood Pressure in Humans: A Scientific Statement From the American Heart Association.” Hypertension, 2019; 73(5): e35-e66. https://pubmed.ncbi.nlm.nih.gov/30827125/ ↩ ↩²
Staplin N, de la Sierra A, Ruilope LM, Emberson JR, Vinyoles E, Gorostidi M, Ruiz-Hurtado G, Segura J, Baigent C, Williams B. “Relationship between clinic and ambulatory blood pressure and mortality: an observational cohort study in 59 124 patients.” The Lancet, 2023; 401(10393): 2041-2050. https://pubmed.ncbi.nlm.nih.gov/37156250/ ↩ ↩² ↩³
Jones DW, Ferdinand KC, Taler SJ, Johnson HM, Shimbo D, Abdalla M, Altieri MM, Bansal N, Bello NA, Bress AP, Carter J, Cohen JB, Collins KJ, Commodore-Mensah Y, Davis LL, Egan B, Khan SS, Lloyd-Jones DM, Melnyk BM, Mistry EA, Ogunniyi MO, Schott SL, Smith SC Jr, Talbot AW, Vongpatanasin W, Watson KE, Whelton PK, Williamson JD. “2025 AHA/ACC/AANP/AAPA/ABC/ACCP/ACPM/AGS/AMA/ASPC/NMA/PCNA/SGIM Guideline for the Prevention, Detection, Evaluation and Management of High Blood Pressure in Adults.” Hypertension, 2025; 82(10): e212-e316. https://pubmed.ncbi.nlm.nih.gov/40811516/ ↩
Edwards JJ, Deenmamode AHP, Griffiths M, Arnold O, Cooper NJ, Wiles JD, O’Driscoll JM. “Exercise training and resting blood pressure: a large-scale pairwise and network meta-analysis of randomised controlled trials.” British Journal of Sports Medicine, 2023; 57(20): 1317-1326. https://pubmed.ncbi.nlm.nih.gov/37491419/ ↩ ↩² ↩³ ↩⁴