HRV Explained: SDNN vs RMSSD

Your iPhone says your HRV is 48 ms. A friend's ring says theirs is 75 ms. A different app shows you at 31 ms the same morning. None of these numbers has to be wrong, but they may not measure the same thing, and putting them side by side tells you very little.

This guide covers what heart rate variability is, the two metrics you're most likely to see (SDNN and RMSSD), why your own baseline is the only fair comparison, and exactly how Tracked labels the HRV it imports.

What HRV is

A healthy heart doesn't beat like a metronome. Even at rest, the time between one beat and the next changes by a few milliseconds from beat to beat. Heart rate variability (HRV) describes how much those intervals vary.

That variation is shaped largely by the autonomic nervous system, and breathing, posture, sleep and stress all feed into it (Shaffer & Ginsberg, 2017). HRV isn't one number with one formula. It's a family of statistics, each calculated differently from the same series of beat intervals. The two you'll meet most often on consumer devices are SDNN and RMSSD.

SDNN: overall variability across a recording

SDNN is the standard deviation of the normal-to-normal (NN) intervals: the gaps between ordinary heartbeats, with artefacts and irregular beats removed. It describes how spread out all the intervals in a recording are.

Because SDNN captures every kind of variation across the whole recording, slow swings included, it depends heavily on how long the recording is. The 1996 Task Force that standardized HRV measurement advised against comparing SDNN values taken from recordings of different lengths (Task Force, 1996).

RMSSD: beat-to-beat variability

RMSSD is the root mean square of successive differences. Take the difference between each pair of neighbouring intervals, square those differences, average them, then take the square root. The result focuses on fast, beat-to-beat change.

RMSSD is the main time-domain measure used to estimate vagally mediated (parasympathetic) HRV (Shaffer & Ginsberg, 2017). It also works on short recordings, which is one reason sports scientists commonly use it to monitor athletes (Buchheit, 2014).

Why SDNN and RMSSD aren't interchangeable

• They're different formulas. The same heartbeats give two different numbers, because the two statistics answer different questions about those beats.
• There's no conversion factor. The gap between them isn't fixed. It varies from person to person and from recording to recording.
• Devices measure differently, too. Recording length, time of day, sensor type (chest-strap ECG or an optical sensor on the wrist or finger) and artefact handling all shift HRV values. Methodological differences like these explain much of the spread in published values (Nunan et al., 2010).

So 45 ms SDNN and 45 ms RMSSD are not the same reading. Going from 40 ms SDNN on one device to 55 ms RMSSD on another isn't an improvement, either. It's a change of ruler.

Which platforms report which metric

Some of this is well documented. Where it isn't, check your device maker's own documentation.

• Apple Health stores HRV as SDNN. HealthKit's heart rate variability type is defined as SDNN, and that's what Apple Watch writes.
• Android's Health Connect stores HRV as RMSSD. Its HRV record type is HeartRateVariabilityRmssdRecord.
• Many dedicated wearables report RMSSD. WHOOP and Oura, for example, describe their HRV as RMSSD. Other devices vary, and a maker can change its method over time.

Compare with your own baseline, not with other people

HRV differs enormously between people. A systematic review of short-term HRV covering 44 studies and more than 21,000 healthy adults found large differences between individuals (Nunan et al., 2010). Age, sex and measurement context all shift typical values, and norms for different recording lengths aren't interchangeable (Shaffer & Ginsberg, 2017).

A value that's low for one person can be ordinary for another. The useful question isn't "is 48 ms good?" It's "how does this compare with my own recent normal, measured the same way?" Researchers who monitor elite athletes make the same point: you need longitudinal data to learn an individual's own HRV pattern (Plews et al., 2013).

Why a single night is noisy

HRV moves from day to day for many reasons. Sleep, illness, stress, alcohol, the time a reading was taken and simple measurement noise can all play a part. A single reading can't tell you which, if any, of these was involved.

That's why sports-science monitoring leans on averages rather than single days. Plews and colleagues tracked a 7-day rolling average of RMSSD in elite triathletes (Plews et al., 2012) and recommend appropriate averaging for day-to-day monitoring (Plews et al., 2013). Treat one night as one data point. A week compared with the weeks before it is a steadier picture, though it's still a description, not an explanation.

HRV is not a diagnosis

Consumer HRV is a wellness measurement, not a medical test. A lower number doesn't mean something is wrong, a higher one doesn't mean you're healthy, and neither tells you why it moved. If you have symptoms or concerns about your health, talk to a doctor rather than relying on a wearable reading.

How Tracked labels HRV

When you connect Apple Health or Health Connect through Health Sync, Tracked keeps the metric your source actually reported:

• Apple Health → SDNN.
• Health Connect → RMSSD.
• Wearable account connections → RMSSD. Any HRV Tracked imports directly from a connected wearable account is stored as RMSSD.

A reading that doesn't say which metric it is gets dropped, not guessed. SDNN and RMSSD are stored as two separate metrics, and Tracked never averages them together, converts one into the other, or compares one with the other. If your history contains both (say you moved from an iPhone to an Android phone), Tracked uses one metric for a comparison and leaves the other out instead of blending them. The readiness screen names the metric it used, for example "HRV (SDNN)".

The baseline is yours alone. Tracked compares your last 7 days with the 28 days immediately before them, so the week being judged is never part of its own baseline. It's never compared with a population average or a clinical threshold.

• Enough data first. A change is only reported once the recent week has at least 4 recorded days and the baseline window has at least 14. Until then, Tracked says there isn't enough data to compare. That's different from a change of zero.
• A device switch is flagged, not compared. If none of the sources in your recent week also appear in your baseline window, the two averages come from different instruments. Tracked says the source changed instead of reporting a number. As days from your current source move into the baseline window, the comparison becomes available again.

What Tracked doesn't calculate. Tracked computes no recovery or strain score, and it doesn't turn HRV into a score of its own. The readiness score in Tracked comes from your answers to the readiness survey. With Tracked PRO readiness surveys, HRV and resting heart rate measured against your own baseline can add a small, capped adjustment to that score once there's enough history, and the readiness screen shows how much they moved it. Your survey answers still drive the score.

Asking Claude about your HRV

Tracked's connector is listed in Claude's connector directory and reads the same data, read-only. Its recovery summary names the HRV metric (SDNN or RMSSD) next to every HRV figure and states the baseline window: the last 7 days against the 28 days before them, with how many recorded days sit behind each average. When a comparison isn't possible, it says why. It reports no readiness, recovery or strain score, and it isn't medical advice. See AI Connectors for setup and what else the connector can read.

References

1. 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, 93(5), 1043-1065. PubMed
2. Shaffer, F. & Ginsberg, J.P. (2017). An overview of heart rate variability metrics and norms. Frontiers in Public Health, 5, 258. PMC
3. Buchheit, M. (2014). Monitoring training status with HR measures: do all roads lead to Rome? Frontiers in Physiology, 5, 73. PMC
4. Nunan, D., Sandercock, G.R. & Brodie, D.A. (2010). A quantitative systematic review of normal values for short-term heart rate variability in healthy adults. Pacing and Clinical Electrophysiology, 33(11), 1407-1417. PubMed
5. Plews, D.J., Laursen, P.B., Kilding, A.E. & Buchheit, M. (2012). Heart rate variability in elite triathletes, is variation in variability the key to effective training? A case comparison. European Journal of Applied Physiology, 112(11), 3729-3741. PubMed
6. Plews, D.J., Laursen, P.B., Stanley, J., Kilding, A.E. & Buchheit, M. (2013). Training adaptation and heart rate variability in elite endurance athletes: opening the door to effective monitoring. Sports Medicine, 43(9), 773-781. PubMed
7. Apple Developer Documentation. heartRateVariabilitySDNN. Apple
8. Android Developers. HeartRateVariabilityRmssdRecord. Android