Short answer: HRV declines with age, and that part is well established. But almost every "normal HRV by age" table circulating online traces back to a single 1998 study of 24-hour Holter recordings, using metrics and a time window that have almost nothing in common with the overnight number your ring or strap reports. Comparing your reading against those tables is comparing two different measurements.
The one benchmark that survives scrutiny is narrower, and it is at the end of this piece.
What is actually being measured
Heart rate variability is the variation in time between consecutive heartbeats. A healthy heart is not a metronome, and the beat-to-beat wobble reflects the balance between the sympathetic and parasympathetic branches of your autonomic nervous system.
The measurement standards were set by a joint Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology in 1996, and they still govern the field. Two time-domain metrics matter here.
SDNN is the standard deviation of all normal beat intervals. It estimates overall variability and is meaningfully interpretable over a full 24 hours.
RMSSD is the square root of the mean of squared differences between adjacent intervals. It estimates the short-term component, is closely tied to vagal activity, and is what the Task Force recommends for short recordings.
That distinction is the source of the whole problem below. Consumer wearables report RMSSD, measured overnight. The famous age tables report SDNN and related metrics, measured across a full day and night. They are different numbers describing different windows.
Where the age tables come from
The source is Umetani and colleagues (1998), published in the Journal of the American College of Cardiology. It studied 260 healthy people aged 10 to 99 using 24-hour Holter monitors, stratified by decade, and it is a good study.
What it found is that the decline depends heavily on which metric you look at:
| Metric | What happened with age |
|---|---|
| SDNN and SDANN | Fell gradually to about 60% of baseline by the tenth decade |
| SDNN index | Fell roughly linearly to about 46% |
| RMSSD | Fell to about 47% of baseline by the sixth decade, then stabilised |
| pNN50 | Fell to about 24% of baseline by the sixth decade, then stabilised |
| Sex difference | Women lower than men under 30, narrowing after 30, gone after 50 |
Two things follow that most articles quoting this study omit.
First, the headline results are proportions of a baseline, not a tidy list of millisecond values by decade. Any ms-by-age table you find attributed to this paper has been reconstructed by someone else, and reconstructions vary.
Second, and more important: these are 24-hour Holter values from clinical-grade electrodes. Your wearable measures a subset of the night, from the wrist or finger, with an optical sensor, and reports a different metric. The numbers are not interchangeable, and nothing in that paper licenses comparing them.
The tables are not wrong. They are answers to a different question than the one your ring is answering.
What wearables can and cannot measure
The good news is more encouraging than the houseplant literature.
Cao and colleagues (2022) compared Oura Ring overnight readings against a research ECG in 35 healthy adults sleeping at home. Heart rate and RMSSD were accurate in both five-minute and whole-night analyses. SDNN, pNN50 and average interval were acceptable only when averaged across the whole night. The frequency-domain measures, LF and the LF:HF ratio, had high error in both analyses.
A systematic review by Georgiou and colleagues (2018), covering 18 studies, found agreement with ECG ranging from very good to excellent at rest, declining progressively as exercise intensity increased.
So the device is not the weak link. RMSSD overnight is a genuinely reasonable measurement. The weak link is the benchmark you are comparing it against.
The one number that is defensible
If you want a population reference for the metric your device actually reports, use this one.
Shaffer and Ginsberg (2017) compiled normative values across 15 studies of healthy adults. For short-term recordings, RMSSD averaged 42 ms, with a standard deviation of 15 ms and a range of roughly 19 to 75 ms.
That range is the honest headline. It is enormous. A perfectly healthy adult can sit anywhere across it, which is precisely why a single reading tells you almost nothing about your health relative to other people.
How to use the number properly
Compare yourself to yourself. Your seven-day rolling average against your own last three months is the only comparison that carries information. Absolute values between people are dominated by genetics, body size, resting heart rate and breathing pattern.
Measure under identical conditions. HRV moves with body position, time since waking, breathing rate, hydration and room temperature. Overnight readings from a device you wear every night are the most consistent option available, which is the real argument for wearing one.
Ignore single days. Day-to-day scatter is large and mostly noise. A drop after one bad night means very little. A downward two-week trend means something.
Ignore the LF:HF ratio. Apps present it as a sympathetic-parasympathetic balance score. It is the least reliable thing your device outputs and the interpretation is contested even in clinical research.
Know what reliably moves it. Alcohol is the clearest single input. Pietilä and colleagues (2018) studied 4,098 Finnish employees and found that during the first three hours of sleep, an HRV-derived recovery measure fell dose-dependently with intake: by 9.3, 24.0 and 39.2 percentage units for low, moderate and high consumption. Those are units of a composite score rather than raw RMSSD, but the direction and dose-dependence are unambiguous, and they show up on consumer devices.
Illness, poor sleep, hard training and heat also suppress it. Aerobic fitness and consistent sleep raise it over months.
| If you are looking at | Do this |
|---|---|
| One low morning | Nothing. Check again in a week |
| A two-week downward trend | Look at sleep, alcohol, training load, illness |
| Your number versus a friend's | Stop. Different bodies, possibly different devices |
| Your number versus an online age chart | Stop. Different metric, different recording window |
| A high LF:HF "stress" score | Ignore it |
Questions people ask
What is a good HRV by age? There is no defensible age-by-age millisecond table for wearable readings. The tables online come from 24-hour clinical recordings using different metrics. For short-term RMSSD in healthy adults, published values average around 42 ms across a range of roughly 19 to 75 ms, at any adult age.
Is 30 ms HRV bad? Is 90 ms good? Both sit within or near the published healthy range. Neither number alone indicates anything about your health. Your own trend is the signal.
Does HRV decrease with age? Yes. The decline is well documented, though its size depends on which metric is used, and some measures plateau after middle age rather than falling continuously.
Why is my HRV so different from my partner's? Between-person variation is large and largely constitutional. Different devices and different measurement windows widen the gap further.
Which HRV metric should I actually watch? RMSSD, averaged overnight, viewed as a rolling weekly average. It is the vagally-mediated short-term metric, it is what most wearables report, and it is the one that validates well against ECG.
This article covers cardiac physiology for general information and is not medical advice. A sustained unexplained change in HRV, palpitations, or symptoms such as dizziness or chest pain should be assessed by a clinician rather than tracked with an app.
References
- 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. doi:10.1161/01.CIR.93.5.1043
- Umetani, K., Singer, D.H., McCraty, R., & Atkinson, M. (1998). Twenty-four hour time domain heart rate variability and heart rate: relations to age and gender over nine decades. Journal of the American College of Cardiology, 31(3), 593–601. PubMed 9502641
- Shaffer, F., & Ginsberg, J.P. (2017). An overview of heart rate variability metrics and norms. Frontiers in Public Health, 5, 258. doi:10.3389/fpubh.2017.00258
- Cao, R., Azimi, I., Sarhaddi, F., et al. (2022). Accuracy assessment of Oura Ring nocturnal heart rate and heart rate variability in comparison with electrocardiography in time and frequency domains. Journal of Medical Internet Research, 24(1), e27487. doi:10.2196/27487
- Georgiou, K., Larentzakis, A.V., Khamis, N.N., et al. (2018). Can wearable devices accurately measure heart rate variability? A systematic review. Folia Medica, 60(1), 7–20. PubMed 29668452
- Pietilä, J., Helander, E., Korhonen, I., Myllymäki, T., Kujala, U.M., & Lindholm, H. (2018). Acute effect of alcohol intake on cardiovascular autonomic regulation during the first hours of sleep in a large real-world sample of Finnish employees. JMIR Mental Health, 5(1), e23. doi:10.2196/mental.9519
The weekly readout
One email each Thursday: what we tested, which claim collapsed under a closer look, and the one number worth paying attention to.