What is HRV? A plain-language guide to heart rate variability

Heart rate variability is, at the level of physics, very simple. Your heart does not beat like a metronome. The interval between one beat and the next is never exactly the same as the interval before it. That small, beat-to-beat fluctuation, measured in milliseconds, is HRV.

The reason wearables have spent the last six years putting HRV on your phone every morning is that this fluctuation correlates with something useful: the balance between the two branches of your autonomic nervous system. When the parasympathetic side (the “rest and digest” branch, carried mostly by the vagus nerve) is active, beat-to-beat variability goes up. When the sympathetic side takes over, variability goes down. A single number derived from a few minutes of pulse data carries a real, if noisy, signal about how recovered, stressed, or strained you are.

That is the whole pitch. Most of what wearables tell you about HRV is built on that one fact, plus a lot of math and opinion about what to do with the result.

What is actually being measured

The raw quantity behind every HRV metric is the RR interval: the time, in milliseconds, between two consecutive R waves on an ECG. The R wave is the sharp upward spike that corresponds to ventricular contraction. On a clean ECG it is unambiguous, which is why ECG is the reference method for HRV.

A typical RR interval at rest sits between 700 and 1100 ms (a heart rate of roughly 55-85 BPM). In a relaxed adult, successive intervals will be 30, 50, sometimes 80 ms apart from each other. That spread is HRV. Under acute stress, after a hard workout, or with a high resting heart rate, the spread shrinks and the intervals look more uniform.

You do not see this on a heart-rate display. A wearable smoothing pulse rate over five seconds will tell you “72 BPM” whether the underlying intervals are dancing between 800 and 880 or marching at a flat 833. HRV is what comes out when you stop averaging and look at the gaps.

RMSSD, SDNN, and the time-domain math

Once you have a string of RR intervals, you need a single number to put on a screen. The HRV literature has dozens of candidate metrics. Consumer devices have converged on two of them.

The first is RMSSD, the root mean square of successive differences. Take each pair of adjacent RR intervals, find the difference, square it, average across the window, take the square root. RMSSD is dominated by short-term changes, which makes it sensitive to parasympathetic (vagal) tone, and it is relatively stable over short three-to-five-minute windows. Almost every consumer device uses it.

The second is SDNN, the standard deviation of all NN intervals (NN being “normal” RR intervals with ectopic beats filtered out). SDNN captures variability over the full window, including slower oscillations driven by breathing, blood pressure regulation, and circadian effects. It is the metric the Apple Watch reports in the Heart Rate Variability tile, and it needs a longer window to stabilize. A 30-second Apple Watch SDNN reading is genuinely a different animal from a five-minute clinical SDNN.

There is also a frequency-domain family (LF and HF power, and the LF/HF ratio) that comes from a Fourier transform of the RR interval series. The HF band (roughly 0.15-0.40 Hz) reflects parasympathetic activity; the LF band is more contested. Almost no consumer wearable surfaces frequency-domain HRV, because the interpretation is messy and the math is sensitive to artifact. Researchers care; your wrist does not.

For a consumer in 2026, the practical takeaway: if your device reports a single “HRV” number, it is almost certainly RMSSD over a window the manufacturer has picked. Whoop, Oura, and Garmin all use RMSSD. Apple is the outlier, reporting SDNN, which is part of why Apple’s HRV numbers tend to read lower than other devices on the same wrist.

Why higher is generally better

In a healthy person at rest, vagal tone is high. The heart is being actively modulated by the parasympathetic nervous system, including breath-linked oscillations (respiratory sinus arrhythmia, where heart rate rises slightly on the inhale and falls on the exhale). All of that adds beat-to-beat variability and pushes RMSSD up.

Acute stress, intense exercise, alcohol, illness, dehydration, and poor sleep push HRV down by suppressing parasympathetic activity and raising sympathetic activity, making inter-beat intervals more uniform.

So within an individual, on a comparable measurement, a higher HRV reading is usually a better sign than a lower one. That is the basis of every recovery score the wearables industry sells.

The trouble starts when people compare across individuals. RMSSD in healthy adults ranges from the low 20s to well over 100 ms, with strong dependence on age (HRV declines roughly linearly through adulthood), genetics, fitness, sex, and body composition. A 24-year-old endurance athlete with an RMSSD of 95 and a 55-year-old desk worker with an RMSSD of 28 may both be perfectly healthy. Neither number tells you anything about the other person.

The only HRV comparison that means much is you to you, same time of day, same device. The same is true of resting heart rate.

How the measurement actually happens on your wrist

ECG is the reference. Everything else is an estimate that gets compared to ECG. There are four real-world paths a consumer device uses to get to an HRV number.

Chest straps (Polar H10, Garmin HRM-Pro, Wahoo TICKR) sit on the sternum, pick up the heart’s electrical activity through skin electrodes, and produce an ECG-derived RR stream that is, for practical purposes, as accurate as a clinical Holter. Gold standard short of a hospital monitor. Annoying to wear all night, which is why almost nobody uses them for daily HRV.

ECG-capable smartwatches (Apple Watch since Series 4, Withings ScanWatch, several Samsung Galaxy Watch generations) include a single-lead ECG triggered by touching a finger to the crown or bezel. The reading takes 30 seconds, requires you to sit still, and produces a clean RR series. Those values are good. But because they are captured only during a manual reading, you do not get a continuous baseline. The Apple Watch’s continuous HRV numbers, the ones in the Health app whether or not you have run an ECG, come from a different sensor entirely; we cover the distinction in ECG smartwatch explained.

Wrist PPG is the optical sensor on the back of every fitness watch. A green LED shines into the skin, a photodiode measures the light coming back, and the algorithm reconstructs the pulse waveform. From there you can pick out peaks (the “PP intervals,” approximating RR intervals) and compute HRV. This works. Accuracy depends on a clean optical signal: no motion, no sweat sliding, decent skin contact. Wrist PPG is the noisiest of these four methods, but it is the only one that captures HRV continuously across a full day, which is why Whoop, Fitbit, and Garmin rely on it.

Finger PPG is what the Oura Ring uses, and what Ultrahuman and several other rings have copied. Same physics, different anatomy. The finger has thinner skin, higher blood perfusion, and less muscle interference, which makes the optical signal cleaner. Oura computes RMSSD overnight from this signal, and the published validation work, including Stone et al. 2021 in Sensors, puts Oura’s overnight HRV within a few milliseconds of an ECG reference for most participants. Compared head-to-head with wrist PPG, finger PPG wins on overnight HRV almost every time.

We have written more on the tradeoffs between these specific devices in Oura Ring vs Whoop and in the best sleep trackers roundup, both of which treat HRV reporting as one of the main scoring categories.

What ruins an HRV reading

Even from a chest strap, HRV is a delicate signal. From a wrist or a ring, it is fragile.

Motion is the obvious wrecker. Any optical sensor that loses lock on the pulse waveform during a reading will either drop out or fabricate intervals from noise. Wrist PPG during a workout, or even a fidgety hour at a desk, often produces HRV values that are mostly artifact. Most devices refuse to report HRV during motion. The ones that report all-day HRV (Whoop is the leading example) lean heavily on quiet windows and overnight reads.

Alcohol is the second wrecker, and the most reliable demonstration that HRV is a real signal. Two drinks within four hours of bedtime will move overnight RMSSD by a noticeable margin, often 20-30%. Three or more drinks usually flatten it. This is one of the few things you can watch happen on your own device inside 24 hours, and it is part of why HRV has caught on with people who do not otherwise care about cardiac physiology.

The full list of HRV depressants is long and unsurprising: caffeine close to bedtime, late heavy meals, viral infection (often two or three days before any subjective symptom), heavy training the day before, jet lag, a hot bedroom, emotional stress, dehydration.

Sensor confounders add more noise. PPG-derived HRV gets worse with darker skin pigmentation, because green light is more absorbed and less scattered back to the photodiode (Bent et al. 2020, npj Digital Medicine). Tattoos under the sensor produce similar problems. Loose fit, cold extremities, and fine wrist hair all add noise.

What apps do with the number

By the time HRV reaches the screen of a phone, it has been wrapped in a layer of product opinion that often hides the underlying value entirely.

Whoop Recovery is the most aggressive about this. The score is computed nightly from overnight HRV, resting heart rate, respiratory rate, and sleep performance, with HRV dominating the math. Whoop shows the raw RMSSD in the app, but the daily UI is built around a 0-100% percentage that is, in effect, your overnight HRV normalized against your personal baseline. The strength is that it sidesteps the cross-individual comparison problem. The weakness is that on any given morning you do not know how much of the score is HRV versus the other inputs.

Oura Readiness does something similar. It blends HRV, resting heart rate, body temperature, and recent activity into a 0-100 score, and surfaces overnight RMSSD as a standalone trend graph.

Garmin Body Battery uses HRV from wrist PPG as one input in a 0-100 energy-reserve estimate. It is the least HRV-centric of the major recovery scores.

Apple Watch Cardio Recovery is a different beast. It measures how quickly your heart rate falls in the minute after a workout ends, a parasympathetic-reactivation metric related to HRV but computed from heart rate dynamics. Apple’s continuous HRV tile in the Health app reports SDNN from passive readings and does not roll into any unified recovery score.

These are useful the way a thermometer is useful: a daily check that catches deviations from your own baseline. They are not a leaderboard.

What HRV cannot tell you

We have rewritten this section the most, because consumer HRV content drifts toward either over-claiming (your HRV reveals everything about your health) or over-debunking (HRV is meaningless). Both are wrong.

HRV is not a readiness oracle. It correlates with autonomic state, and autonomic state correlates with how you feel. It does not predict whether you should skip a workout. Plenty of athletes have set personal bests on low-HRV mornings, and plenty have crashed on high-HRV ones. If your device tells you to take it easy because HRV is down, treat that as one input, not a verdict.

HRV is not a sleep quality measure on its own. It is correlated with sleep structure, especially time in deep and REM stages, but it is not measuring sleep. A device that uses HRV alongside accelerometry can stage sleep with reasonable accuracy (we cover this in sleep stages explained). The HRV number alone cannot tell you whether you slept well.

HRV is not a fitness improvement measure on its own. Over months, a consistently training person will tend to see resting HRV creep up and resting heart rate creep down. That trend is real, slow, and easily masked by life. Picking up an HRV device expecting to watch fitness improve in four weeks is a recipe for disappointment.

HRV cannot diagnose anything. No consumer wearable is FDA-cleared to use HRV as a diagnostic input. The clinical literature linking very low HRV to elevated cardiovascular risk uses ECG-derived HRV in controlled settings, not your wrist on a Tuesday.

The Basis Peak, for what it is worth, never reported HRV usefully. The optical sensor captured pulse data, but the firmware never exposed HRV as a metric, and the cloud streamed beat-summary data rather than RR intervals. Archived Peak exports cannot retroactively give you an HRV trend. The full sensor picture is in Basis Peak heart-rate accuracy.

Five rules for using HRV without losing your mind

The point of HRV is not to chase a number. The point is a daily marker that catches what your subjective sense missed: an oncoming illness, an under-recovered week, the second drink hitting harder than the first.

1. Establish a baseline before you draw conclusions. Two weeks of consistent overnight readings, ideally three, before you decide what your normal looks like. A single high or low morning means nothing.
2. Trend over absolute, every time. Your number against your last 30 days is the signal. Your number against your friend’s number is noise.
3. Use the same device, same conditions. Switching from a Whoop to an Oura mid-baseline resets your reference. So does switching from overnight to spot readings, or moving the device to a different finger.
4. Treat single nights as noise; treat weeks as data. A 20% drop one morning, after a hard run and a late dinner, is a yawn. A 20% drop sustained for five mornings is worth a closer look.
5. Pair HRV with at least one other signal. Resting heart rate is the easiest second signal, and most devices report both. Subjective how-do-I-feel scoring is the cheapest. The combination catches things HRV alone misses.

If you are picking a device for the first time and HRV is a reason, rings and chest-strap-plus-app combinations currently outperform wrist watches for overnight HRV accuracy; our current view of the field is in best health trackers. The marketing on every recovery-scoring device will tell you their HRV is the best one. Some of them are even right. The number that matters, still, is your own.