How Apple Watch Measures HRV and Why the Number Changes

Twenty people sat in front of a screen for five minutes wearing an Apple Watch and a Polar H7 chest strap, and the two devices agreed on the timing of more than 12,000 heartbeats with a concordance of 0.989. That was Hernando and colleagues' 2018 validation in Sensors. Four years later Miller and colleagues put an Apple Watch Series 6 on 53 sleeping adults next to a laboratory ECG. Across the night the watch read RMSSD 9.6 ms low on average, and its 95 percent limits of agreement stretched 55 ms either side of that average.

Both results are correct. They describe different measurements from the same watch, and most confusion about Apple Watch HRV comes from treating them as one number. The watch records HRV at three kinds of moments, stores it with a formula most sport science does not use, and never labels which moment produced a reading. What follows covers each piece, the validation data, and a protocol for one comparable reading per day.

01How Apple Watch takes an HRV reading

Apple Watch has no button for an HRV test. HealthKit's developer reference says the system "automatically records samples on Apple Watch," and Apple's support page adds that HRV "is sampled while you're still, so the number of measurements you see will vary according to your activity level." The watch waits for a window when your wrist is quiet and the optical signal is clean, records the beats in that window, and writes one value to Apple Health.

The sensor is photoplethysmography. Green LEDs shine into the wrist, photodiodes measure the reflected light, and each pulse of blood shows up as a change in that light. The watch times the milliseconds between pulse peaks, and HRV is a statistic of that interval series. Motion that smears the pulse wave, or a gap where the signal drops, degrades the series before any formula runs.

Heart rate and HRV follow different schedules. Apple says the Series 12 and Ultra 4 can measure heart rate as often as every five seconds, even in motion. HRV still waits for stillness, and Apple describes the Recovery HRV figure on those models the same way. Neither Apple page states how long each HRV recording lasts.

02The three moments HRV gets recorded

Every HRV value in Apple Health comes from one of three situations, and each one captures a different autonomic state.

Background readings during the day. These arrive a few times a day at moments the watch picks, usually while you sit at a desk or drive. Your posture, your last coffee and your breathing rate all ride along in the sample. A background value at 10:40 on a Tuesday and another at 15:15 on a Thursday differ for reasons that have nothing to do with how well you recovered.

Overnight readings. With the watch on in bed, the sensor gets hours of stillness and records several values across the night, with constant posture and unforced breathing. This is the least contaminated state the watch sees all day, which is why Titan's help center names sleep HRV as the steadiest input (Recovery settings). The watch has to be on your wrist and charged, and sleep tracking has to be on in the Health app (Apple Watch support).

Mindfulness sessions. Starting a session in the Mindfulness app records HRV while it runs and ties the reading to the session's timestamp. Before watchOS offered developers direct access to beat-to-beat data, this was the only way to get raw intervals out of the watch. Hernando and colleagues used exactly this route, the Breathe app under watchOS 4.2, to extract the interval series they validated. A session is the one HRV reading you can trigger on demand.

03What Apple Health stores

Apple Health keeps HRV in a data type called heartRateVariabilitySDNN. Apple's reference says it directly. "While there are multiple ways of computing HRV, HealthKit uses SDNN heart rate variability, which uses the standard deviation of the inter-beat (RR) intervals between normal heartbeats." Each sample is one number in milliseconds with a start and end time.

Apple Health can also hold a heartbeat series, the beat-to-beat timing itself, which lets another app compute a different statistic from the same beats. The HRV in the Health app is SDNN, while WHOOP, Oura, Garmin and Polar report RMSSD, so two apps showing different values for the same night are often running two formulas. The wearables and performance data guide compares the devices.

04SDNN and RMSSD are different measurements

Both statistics start from the same list of intervals between normal beats. The 1996 Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology defined them and they have not changed since.

SDNN is the standard deviation of the intervals. Take the mean interval, measure how far each interval sits from it, square those distances, average them and take the square root. It asks how spread out the intervals are around their average.

RMSSD is the root mean square of successive differences. Subtract each interval from the next one, square the differences, average them and take the square root. It asks how much each beat differs from the beat right before it.

Take five intervals of 1,000, 1,050, 980, 1,040 and 990 ms. The mean is 1,012 ms. The deviations from the mean are −12, 38, −32, 28 and −22, their squares sum to 3,880, and dividing by four and taking the root gives an SDNN of 31.1 ms. The successive differences are 50, −70, 60 and −50, their squares average 3,375, and the root gives an RMSSD of 58.1 ms. Same five beats, two answers almost a factor of two apart.

The gap matters because each statistic responds to a different kind of variation. The next table shows two eight-beat series with the same mean interval of 1,000 ms.

SDNN barely tells them apart. RMSSD separates them by a factor of 3.5. Fast beat-to-beat change comes mostly from the vagus nerve, which can alter the heart's timing within a single beat. Slow drifts come from sympathetic tone, blood pressure regulation, temperature and posture. Shaffer and Ginsberg's 2017 review in Frontiers in Public Health calls RMSSD "the primary time-domain measure used to estimate the vagally mediated changes reflected in HRV." SDNN mixes both kinds of variation, so it moves with vagal recovery and with everything else.

SDNN also grows with recording length, because a longer window catches more slow drift. The Task Force warns against comparing SDNN from recordings of different durations. Shaffer and Ginsberg add that in short resting recordings the main source of SDNN variation is respiratory sinus arrhythmia, the vagal swing tied to breathing, "especially with slow, paced breathing." In still, short recordings the two move together. In a free-living background reading, SDNN picks up more slow noise.

05Why a Mindfulness reading runs higher

Heart rate rises on each inhale and falls on each exhale. Slow, deep breathing widens that swing, and near six breaths per minute it grows close to its maximum, which is the basis of HRV biofeedback. Both SDNN and RMSSD grow with the swing, so a Mindfulness session where you follow a slow pacing cue records how large an oscillation you can produce on command, a different quantity from the HRV of unguided breathing at rest. The mindfulness and breathwork guide walks through the evidence and shows how one paced-breathing session can move Titan's Recovery and Stress scores on a morning when your physiology has not changed. If you use a morning session as your daily reading, breathe at your natural rate and ignore any pacing animation.

06Why the number jumps around

HRV moves from day to day even when measured perfectly. Sleep, alcohol, heat, illness onset, the menstrual cycle and yesterday's training all shift it. Plews and colleagues, writing in Sports Medicine in 2013 about elite endurance athletes, worked through the problems of using HRV as a day-to-day monitoring tool and recommended appropriate averaging techniques in place of single readings.

Apple Watch adds its own variation on top. Background readings land at unpredictable times, and HRV follows a circadian pattern, so readings hours apart are not comparable. A seated reading after a meeting and a supine one at 03:00 capture different autonomic states. A shifted strap or a moment of wrist movement can drop beats from a short window.

This is why the number you want is a daily value built the same way every day. Titan compares that value with the median of your own history in units of your typical day-to-day spread, the median absolute deviation, so a 5 ms change counts for more if your HRV rarely moves (Recovery score). One rule of thumb holds outside the app too. A morning HRV more than one median absolute deviation below your baseline median for two or more consecutive days, with resting heart rate above baseline, is a signal to pull intensity. The baselines article covers how that median and spread are built. If you want today's Recovery to stay fixed as later readings arrive, Enable Titan to Save Resting Heart Rates keeps the first valid score Titan saves for the day; it does not write resting heart rate samples to Apple Health (Recovery settings).

07What a typical number looks like

The most common question about Apple Watch HRV is what counts as a good number. The best available reference is a pooled table from Nunan, Sandercock and Brodie's 2010 systematic review in Pacing and Clinical Electrophysiology, which gathered 44 studies and 21,438 healthy adults. Shaffer and Ginsberg reproduce it as follows.

These are short, controlled resting recordings from laboratory ECG. Apple Watch SDNN comes from shorter optical windows taken at uncontrolled moments, so the table gives a sense of scale and nothing more. Nunan's group also found very large differences between individuals and between labs, driven partly by how the recordings were cleaned.

You will also see SDNN cut points quoted online, below 50 ms unhealthy, 50 to 100 ms compromised and above 100 ms healthy. Shaffer and Ginsberg note that those come from 24-hour monitoring, which captures a full day of circadian drift. The Task Force warning about recording length applies, and a watch reading of 40 ms says nothing about which band you belong in. Age moves HRV more than almost any other factor, and the HRV by age article covers what the population data show across decades.

08Accuracy against a chest strap

In Hernando and colleagues' validation, 20 healthy volunteers sat at a monitor wearing an Apple Watch and a Polar H7 chest strap. They watched a relaxing video for five minutes, rested one minute, then did five minutes of an online Stroop test, naming the ink color of mismatched color words. The watch's interval series came from the Breathe app.

Across 12,109 paired intervals, the mean difference between watch and strap was 0.06 ms, and 96.2 percent of pairs fell within the limits of agreement. The concordance correlation was 0.989 at rest and 0.977 under stress.

The watch's recordings had gaps. There were 206 in total, about five per recording, averaging 6.5 seconds each, and about 10 percent of intervals were missing. Those gaps did not produce significant differences in SDNN, RMSSD or pNN50 compared with the strap. Both devices showed RMSSD and high-frequency power falling during the Stroop test, so the watch picked up a real autonomic response to mild mental stress.

Twenty healthy people, seated and still, five-minute recordings through a breathing app. That describes a Mindfulness session closely. It says nothing about background samples, readings across a night, or people with arrhythmias.

09Accuracy across a night

Miller, Sargent and Roach tested six devices, including an Apple Watch Series 6, on 53 adults who spent one night in a sleep laboratory with electrocardiography as the reference. Apple Watch sleep and HRV data were extracted with Sleep Watch, a third-party app, so the Apple result reflects that app's processing of watch data. Garmin HRV came from a three-minute manual test while awake, and the WHOOP and Somfit data were supplied by the manufacturers.

The Apple Watch bias was proportional. The watch overestimated HRV in people with low ECG values and underestimated it in people with high values, so the error depends on where you sit. The authors rated the Apple Watch agreement as good.

For contrast, Kinnunen and colleagues at Oura compared an Oura ring with ECG in 49 adults in Physiological Measurement in 2020 and found nightly RMSSD agreement of r² = 0.980 with a mean bias of −1.2 ms. Those numbers belong to a finger sensor and do not transfer to Apple Watch. They show that a long, still recording gives an optical sensor its best chance of matching ECG.

10What the studies leave open

No published study has tested the SDNN values the watch writes to Apple Health on its own schedule, in free-living conditions, against a simultaneous ECG. Hernando validated a seated five-minute session and Miller validated one night through a third-party app. The daytime background readings that fill most people's Health charts have the thinnest evidence behind them.

Motion is the best-documented source of error for wrist optical sensors. Bent and colleagues tested consumer and research wearables in npj Digital Medicine in 2020 and found that absolute heart rate error during activity was on average 30 percent higher than at rest. They found no statistically significant difference across skin tones. That study measured heart rate, and HRV depends on timing every beat correctly, so the same errors carry over.

11Watch fit and settings that change your reading

Apple's guidance on accurate measurements tells you to wear the watch "snugly on top of your wrist," with the sensor close to the skin. A band loose enough to slide lets ambient light in and lets the sensor move against the skin with every wrist turn, which drops beats.

Blood flow in the skin matters too. Apple notes that skin perfusion varies between people and with the environment, and that in the cold the perfusion at the wrist can fall too low for the sensor to get a reading. Tattoos are the other documented problem. Apple says the ink, pattern and saturation of some tattoos can block the sensor's light. If your readings are sparse and you have ink under the watch, wear it on the other wrist.

If you also wear an Oura ring connected to Titan, two devices may write HRV to Health. Open HealthKit Sources, under Settings > Health & Connected Apps, and set Apple Watch as the Primary Source for HRV so your baseline comes from one sensor (data sources).

12Why HRV can be missing entirely

A blank HRV chart usually has a mundane cause. Titan's HRV not showing article lists five checks, in order.

1. Wear the watch to bed. A night of sleep gives it the best odds of a reading, and a day with no reading at all leaves HRV blank.
2. Run a Mindfulness session before noon. A one-minute session right after waking covers the mornings the watch missed a reading overnight.
3. Check the RMSSD setting. With Use RMSSD for HRV on, Titan needs read access to Heartbeat Series and does not fall back to SDNN.
4. Check your source filters. If the HRV row under HealthKit Sources names a device instead of All Sources, a new watch stays excluded until you allow it. The Mindful Minutes row filters sessions the same way.
5. Confirm Health permission for Heart Rate Variability, Heartbeat Series and Mindful Minutes. iOS never tells an app which read permissions you denied, so Titan cannot warn you.

13How to get one consistent reading each day

The goal is a number taken in the same state at the same point in your day, so that change in the number reflects change in you. Two routes work on Apple Watch.

The first is sleep. Charge the watch before bed, wear it snug, and let it collect readings across the night. Titan averages every HRV reading inside your sleep session, which smooths the spread of individual samples. This route needs no effort each morning, and it is the one to prefer if you meditate or do breathwork in the morning for its own sake.

The second is a short morning session. Right after waking, before caffeine and before getting up, start a one-minute Mindfulness session. Stay in the same position every day, keep your wrist still and breathe at your natural rate. This gives you a controlled reading on nights the watch was on the charger, and it is the closest thing on Apple Watch to the morning supine reading sport scientists use. The HRV and training readiness guide covers that protocol and how to act on it.

Pick one route and stay with it. A baseline built from a mix of sleep averages on some days and paced-breathing sessions on others mixes two measurements, and the median and spread stop meaning anything. The same goes for SDNN and RMSSD. Whichever formula you choose, keep it for the life of the baseline.

14How Titan picks one HRV value per day

Titan uses one HRV value per day for Recovery. With Prioritize Sleep or Mindfulness HRV on, the default, it checks three sources in order and takes the first that has data (Recovery settings).

1. HRV from a Mindfulness session that started before noon. If you logged more than one that morning, Titan uses the most recent session with an HRV reading.
2. The average of all HRV readings during your night's sleep session.
3. The average of all HRV from the last 24 hours, or that calendar day for past days.

The order matches the three moments the watch records HRV, ranked by how controlled they are. For today, Titan looks at Mindfulness sessions only until noon and then moves to sleep HRV, so a morning Recovery built from a session can change after 12:00. Turn the toggle off and Titan skips to the all-day average, which moves through the day as new background readings arrive.

Ignore Mindfulness App HRV, off by default, removes the first step and also drops session readings from the all-day fallback. Turn it on if your sessions are practice and your measurement is sleep.

Use RMSSD for HRV, also off by default, has Titan compute RMSSD from the Heartbeat Series in place of Apple's SDNN. Titan breaks the calculation at any gap so missed beats do not inflate the value, and each series needs at least three beats. Hernando's data show why. About 10 percent of the watch's intervals were missing, and a difference taken across a gap compares two beats that were never neighbors. The setting covers Recovery, Stress and Battery on your iPhone, loads more slowly and needs Heartbeat Series permission. After switching, tap Clear Recovery Cache under Settings > Health & Connected Apps > Resync & Troubleshooting so your history uses one scale. The Titan watch app always uses SDNN.

Recovery then compares the daily value with the median of your 7, 30 or 60 day baseline window, 60 by default, once the window holds at least 7 days covering 60 percent of its days (baselines).

15Set your HRV source in Recovery Preferences

Every choice in this article comes down to one screen. Open Settings from the avatar on Today, then You > Settings, and go to Settings > Recovery & Sleep > Recovery Preferences. Prioritize Sleep or Mindfulness HRV and Ignore Mindfulness App HRV decide which of the watch's three kinds of reading becomes your daily HRV. Use RMSSD for HRV decides whether Titan scores it as Apple's SDNN or as RMSSD from your beat-to-beat data. Set them once to match the protocol you chose, then leave them alone for the life of your baseline. Recovery settings documents each toggle.

16References

1. Apple. heartRateVariabilitySDNN. HealthKit developer documentation. https://developer.apple.com/documentation/healthkit/hkquantitytypeidentifier/heartratevariabilitysdnn
2. Apple. Monitor your heart rate with Apple Watch. Apple Support. https://support.apple.com/en-us/HT204666
3. Apple. Get the most accurate measurements using your Apple Watch. Apple Support. https://support.apple.com/en-us/105002
4. Bent B et al. (2020). Investigating sources of inaccuracy in wearable optical heart rate sensors. npj Digital Medicine. https://doi.org/10.1038/s41746-020-0226-6
5. Hernando D et al. (2018). Validation of the Apple Watch for heart rate variability measurements during relax and mental stress in healthy subjects. Sensors. https://doi.org/10.3390/s18082619
6. Kinnunen H et al. (2020). Feasible assessment of recovery and cardiovascular health: accuracy of nocturnal HR and HRV assessed via ring PPG in comparison to medical grade ECG. Physiological Measurement. https://doi.org/10.1088/1361-6579/ab840a
7. Miller DJ et al. (2022). A validation of six wearable devices for estimating sleep, heart rate and heart rate variability in healthy adults. Sensors. https://doi.org/10.3390/s22166317
8. Nunan D et al. (2010). A quantitative systematic review of normal values for short-term heart rate variability in healthy adults. Pacing and Clinical Electrophysiology. https://doi.org/10.1111/j.1540-8159.2010.02841.x
9. Plews DJ et al. (2013). Training adaptation and heart rate variability in elite endurance athletes: opening the door to effective monitoring. Sports Medicine. https://doi.org/10.1007/s40279-013-0071-8
10. Shaffer F et al. (2017). An overview of heart rate variability metrics and norms. Frontiers in Public Health. https://doi.org/10.3389/fpubh.2017.00258
11. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996). Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Circulation. https://doi.org/10.1161/01.CIR.93.5.1043