Apple Watch Hypertension Alerts: What They Can and Cannot Tell You

Last updated: October 2026 · Based on Apple's FDA-cleared validation data (September 2025) and the peer-reviewed JAMA analysis published in February 2026.

Your watch cannot measure your blood pressure. It has no cuff, it never squeezes your arm, and it will never show you a systolic number. What newer Apple Watches can do — a feature cleared by the U.S. Food and Drug Administration in September 2025 and rolled out in more than 150 countries — is watch the shape of your pulse in the background for 30 days and tell you if that pattern looks like the patterns seen in people with chronic high blood pressure. That distinction, between a notification and a measurement, is the whole story of this feature.

How a light sensor spots a pressure problem

The watch uses photoplethysmography (PPG): the same green-light optical heart sensor that tracks pulse, shining light into the wrist and reading how blood volume changes with each heartbeat. Stiffer arteries and chronically raised pressure change the pulse waveform in subtle, consistent ways. Apple says its model was developed with data from more than 100,000 participants and then validated in a clinical study of about 2,000 people, according to its September 2025 announcement.

The feature runs passively during waking hours, needs no calibration cuff, and judges each 30-day window as a whole. You either receive a “possible hypertension” notification at the end of a window, or you hear nothing. Apple lists it for Apple Watch Series 9 or later and Ultra 2 or later (not the SE), for adults aged 22 and over who have not already been diagnosed with hypertension, and not for use during pregnancy. It was cleared on September 11, 2025 under FDA 510(k) K250507, in a new device category for machine-learning hypertension notification software.

The two numbers that define the feature

In the FDA-reviewed validation, against home blood-pressure monitoring over 30 days, the feature caught 41.2% of people who had undiagnosed hypertension (sensitivity) and correctly cleared 92.3% of people who did not have it (specificity). Read those plainly:

• An alert means something. A test that rarely fires falsely — about 7.7% of people without hypertension received an inappropriate alert — makes an alert worth taking seriously.
• No alert means much less. Because the feature missed about 58.8% of people with hypertension in validation, silence is weak reassurance. Most cases, in that study, produced no notification at all.

Those are unusual proportions for a consumer health feature, and they are the opposite of how most people instinctively read a health app, where “no warning” feels like “all clear.” Here, the warning is the informative part.

What the 2026 JAMA analysis added

In February 2026, researchers at the University of Pennsylvania and the University of Utah published a research letter in JAMA applying Apple's published sensitivity and specificity to nationally representative U.S. blood-pressure data. They estimated a positive predictive value of about 69.1% — roughly seven in ten people who receive an alert actually have hypertension — and a negative predictive value of about 79%, meaning about eight in ten who receive no alert do not have hypertension. Those population figures move with baseline risk, which the authors illustrated by age: in adults under 30, an alert raised the estimated probability of hypertension from 14% to 47%, while no alert lowered it only to 10%. In adults 60 and over, an alert raised it from 45% to 81%, while no alert lowered it to 34% — still a one-in-three chance, which is why the authors stressed that routine cuff measurement remains essential, especially for older and higher-risk adults.

The study did not re-test the watch itself; it modelled what the published test performance would mean across a real population, and it found performance varied by age and by race and ethnicity. That is a reason for clinicians to interpret alerts in context, not a reason to ignore one.

What to do with an alert — and without one

If you get an alert: treat it as a prompt to measure properly, not as a diagnosis. The intended follow-up is confirmation with a validated cuff-based device — home monitoring over several days, following the cuff maker's and your clinician's guidance on seated, rested readings — and a conversation with a clinician about the logged results. A single cuff reading on a stressful afternoon settles little either way.

If you never get an alert: keep whatever screening routine your age and risk already call for. The watch has not cleared you; in validation it missed most cases. Anyone with risk factors, symptoms, or a clinician's advice to monitor should be using a cuff on that advice regardless of what a watch does or does not say.

If you already have a hypertension diagnosis: this feature was not built or cleared for you. Management belongs to cuff readings and your care plan, not to a notification designed for the undiagnosed.

Where this sits among watch health features

Hypertension notifications join a shelf of watch features with very different evidence levels: ECG apps that record a real electrical signal, irregular-rhythm notifications that also work from passive pulse data, and wellness estimates that carry no clearance at all. The buying mistake is treating them as one bundle called “health tracking.” Our AI smartwatches guide separates cleared medical features from wellness features across brands, and our first watch guide puts health features in their proper place behind fit, battery life and the phone you actually own.

There is also a mechanical-watch perspective worth keeping. A traditional watch makes one health-adjacent promise — it tells the time reliably for decades, as our explainer on water resistance ratings and our quartz vs automatic comparison both frame in durability terms. A smartwatch health feature is a screening layer with a battery life, a software support window, and published error rates. Both can be worth owning. Only one of them should ever be read as medical evidence, and even then, only as a first step toward a cuff.

Sources

• U.S. FDA 510(k) K250507 and Apple's September 2025 hypertension notifications validation paper: 30-day PPG windows, 41.2% sensitivity, 92.3% specificity, supported models and intended population.
• JAMA, February 2026 (University of Pennsylvania / University of Utah research letter): population modelling giving ~69.1% positive and ~79% negative predictive value, with age-stratified estimates quoted above.
• Apple Newsroom, September 2025: feature introduction, development data from over 100,000 participants, validation study of about 2,000 participants, availability in 150+ countries.

The single most important takeaway: an alert is a strong reason to reach for a blood-pressure cuff, but no alert is not a clean bill of health — this feature is built to catch some hidden cases, and in validation it missed most of them.

Next steps: read our AI smartwatches guide for how health features compare, or start with our first watch guide for sizing and buying safely.

© FUNGWA 峯華 LLC (fēng huá — “peaks of splendor”)