Heart Rate Variability
The tiny, constantly shifting gaps between your heartbeats are a genuine window on how well your nervous system is coping, and a low reading is consistently linked to higher risk of heart disease and earlier death. But it is a warning light, not a dial: nobody has shown that pushing the number up makes you live longer, and most of what wearables and wellness content claim on top of that is oversold.
You don't need to track HRV at all. If you want to, measure it overnight with a well-validated device, compare each reading only with your own rolling baseline over one to two weeks, and treat a drop that lasts more than a few days as a prompt to look at sleep, alcohol, illness, stress and training load. Fix those upstream causes; don't chase the number.
What the number actually measures
A heart rate of 72 beats per minute is an average. Look at individual beats and the gaps keep shifting — 800 milliseconds, then 850, then 760. That irregularity is healthy, and it comes from two systems pulling against each other. The "fight or flight" side speeds the heart slowly, because its messenger chemical lingers for seconds. The "rest and digest" side, working through the vagus nerve, slows it within milliseconds. So the fast, beat-to-beat wobble is close to a pure read on vagal activity. A heart that ticks like a metronome is one whose vagal brake has weakened — which is what stress, illness, fatigue, depression and aging look like.
HRV measures how your heart rate varies, not how fast it is. The rate is a separate, better-studied number with its own page: Resting heart rate.
Three terms are worth knowing:
- RMSSD is the workhorse most apps report. It tracks the fast vagal changes and is reliable from recordings as short as 10–60 seconds.[1] If you track one number, track this one.
- SDNN measures total variability. It grows with recording length, so a 5-minute SDNN and a 24-hour SDNN are not comparable numbers, and neither are norms built on different recording lengths.
- The LF/HF ratio, still shown in some apps as your "stress balance", does not mean that. Its low-frequency part carries a lot of vagal signal, and things that clearly raise fight-or-flight activity, such as exercise, don't reliably raise it.[2][3] Ignore it.
These definitions come from the field's founding standards document.[4]
Low HRV is a warning, not a lever
That low HRV predicts heart disease and earlier death is well established, from large studies that followed people for years. In the Framingham Heart Study, 736 adults with an average age of 72, each step down in one band of HRV (a mixed signal rather than a pure vagal one) came with about 70% higher risk of dying over four years, after allowing for the usual risk factors.[5] A large US heart study found that low HRV from a two-minute heart tracing predicted coronary disease and death.[6] Pooling 37 groups and about 38,000 people, those with the lowest short-term RMSSD had roughly 56% higher risk of death from any cause.[7]
Three things stop that from meaning "raise your HRV and live longer".
Heart rate muddies it. A faster heart leaves less room for variation, so an uncorrected HRV reading partly just reflects heart rate, and much older research didn't correct for this. In a German study of 1,390 adults followed for 22 years, resting heart rate predicted death from heart disease while none of 66 HRV measures did across the whole group.[8] Heart-rate-corrected HRV did still predict death from any cause in the pooled data above.
The arrow can run backwards. High blood sugar damages the nerves that generate HRV, so diabetes lowers HRV. The reverse is far less clear: in a Dutch study of 7,630 adults, a faster heart rate predicted new type 2 diabetes but corrected RMSSD did not point the same way, and a genetic analysis found no causal link either way.[9]
There are no outcome trials, and genetics don't help. No adequately powered trial has shown that raising HRV prevents heart attacks or death. Studies that use inherited gene variants as a natural experiment found no protective effect of genetically higher HRV on heart attack, heart failure or cardiac death — genetically higher vagal HRV was even linked to more cardiac arrest[10] — and genetically higher SDNN went with about 1.6 times the odds of atrial fibrillation.[11]
Age matters more than any target. HRV declines across life. In a study of 260 healthy people aged 10 to 99, RMSSD had fallen to about half its youthful level by the sixth decade, then levelled off.[12] Large population studies of short resting heart tracings put typical RMSSD at roughly 40–50 ms in your 20s and about 15–20 ms after 60, with a huge spread between healthy people of the same age — roughly 10 to 180 ms at 25–29.[13][14] Overnight wearable readings are measured differently, so treat these as background, not a target.
An "aging clock"? Not yet
HRV is often sold as a cheap biological-age readout. The evidence is early.
In a small study comparing young adults, people in their 80s and centenarians, vagal HRV fell with age, with no significant further drop between the 80s and 100s.[15] An older study of 344 healthy people found vagal HRV lowest in the eighth decade and higher again in the very oldest, hinting that preserved autonomic function may go with healthy longevity.[16] Among the 14 centenarians in that first study who were followed to death, the 4 with the lowest SDNN (below 19 ms) had nearly six times the risk of dying within a year of the other 10 — a striking result from a tiny sample, using a cut-off chosen from the same data.[17]
A 2025 machine-learning study of about 1,060 healthy adults built an "autonomic age" from 29 heart and blood-pressure features: people at high cardiovascular risk came out about 11 years older than their calendar age, and low-risk people about a year younger.[18] It used laboratory equipment, not a wearable, came from one group measured once, and has not been tested against future health.
Claims that HRV tracks the other aging markers are thinner still. Fitter people have modestly longer telomeres, but that study never measured HRV.[19] Almost no studies have measured HRV and epigenetic age in the same people, and the idea that HRV reflects mitochondrial health is a hypothesis.[20] HRV is the easiest aging marker to track at scale, not a master clock.
Inflammation and mood: real links, looser than claimed
The vagus nerve senses inflammatory signals in the blood and, through a reflex loop, can tell immune cells to make fewer of them. That brake is real, but a high HRV reading is a correlate of it, not proof it is working. The best human test is stimulating the nerve directly. In a sham-controlled trial of an implanted stimulator in 242 people with hard-to-treat rheumatoid arthritis, run by the device maker, 35% improved meaningfully against 24% on sham.[21] A non-invasive ear stimulator did no better than sham.[22]
Across 51 observational studies, higher HRV generally went with lower inflammation markers, though some found the opposite.[23] For wearable HRV the picture is weaker: a review of 11 studies found one fairly consistent signal, lower SDNN when the inflammation marker CRP is raised, while links with individual inflammatory molecules were inconsistent.[24] Your watch is not an inflammation test.
Mood follows the same pattern. Across 56 long-term studies, people with lower HRV went on to have slightly more depressive symptoms — a small effect that makes it a vulnerability signal, not a diagnostic test.[25] Antidepressants themselves shift HRV.[26] HRV tends to fall during acute stress, such as surgeons operating, though those studies are small and inconsistent.[27] Ongoing work stress goes with lower vagal HRV, but almost all the evidence is one-off snapshots, and nobody has shown that a falling morning reading can flag burnout before symptoms appear.[28] For what to do about chronic stress, see Stress.
Training by your HRV: same gains, sometimes for less
HRV-guided training is the most practical use of a nightly HRV reading for people who train. Each morning, compare your overnight HRV with your rolling baseline. If it is normal or higher, train as planned. If it is clearly suppressed — typically more than half to one standard deviation below baseline — make the day easy or rest.
The evidence is consistent but modest. Pooling 8 trials, HRV-guided endurance training raised resting vagal HRV more than a fixed plan did, but its edge for fitness was small and not statistically certain; in the trials that reported individual results, fewer people ended up worse off, and some earlier trials got similar gains with fewer hard sessions.[29] A review of strength training and a small randomised trial in cardiac rehabilitation found the same pattern: similar gains, sometimes with less total training.[30][31] No trial has tested whether it prevents injury or overtraining. Act on patterns over 3–7 days rather than on a single morning. For the wider training picture, see VO₂ max and Zone 2.
What actually moves your number
Sleep. Sleep is both the best time to measure HRV and its biggest driver: the vagal brake peaks in deep sleep. Anything that breaks up deep sleep — alcohol, late eating, untreated sleep apnoea, worry — shows up as a lower reading next morning. Treat the sleep, not the number.
Alcohol. In a laboratory study of 26 healthy adults, even one drink for women or two for men sped up the heart for the first hours of sleep; three drinks for women or four for men sped it up more, for most of the night, and more clearly suppressed HRV.[32] It is one of the clearer signals you are likely to see in your own data.
Slow breathing. Breathing at about 6 breaths a minute (roughly 5 seconds in, 5 out), known as resonance breathing, produces large rises in HRV while you do it, and regular practice nudges baseline RMSSD up a little, though most studies are low quality.[33] Doing it with a live HRV display — biofeedback — has moderate evidence for reducing stress and anxiety,[34] and it raises HRV modestly too, though in studies of biofeedback at home the stress benefit wasn't significant.[35]
Meditation hasn't been shown to do the same. Across 19 randomised trials, mindfulness was not significantly better than control for resting vagal HRV.[36] If HRV is the goal, breathe slowly rather than sit still.
Exercise — mixed. Training raised RMSSD clearly in pooled trials of healthy adults[37] and older adults,[38] but a larger synthesis of 34 long-term studies found essentially no change.[39] Exercise is worth doing for many other reasons; don't expect it to reliably move this number.
Diet. A Mediterranean-style diet went with higher HRV in a twin study of 276 middle-aged men,[40] but a small 24-week diet-and-exercise trial didn't improve it.[41] Omega-3 lowers resting heart rate by about 2 beats per minute across 51 trials,[42] but evidence that it raises HRV is weak. It has a catch: in trials of mostly older, high-risk patients, more than 1 g a day raised atrial fibrillation risk by about half, with a smaller, less certain rise at lower doses,[43] and atrial fibrillation falsely inflates HRV. See Omega-3. Habitually eating within a window under 8 hours was linked to more cardiovascular deaths in a large US survey, but that rests on 33 deaths, the people were not deliberately fasting, and confounding can't be ruled out.[44] See Fasting.
Heat and cold. Sauna briefly pushes the nervous system towards fight-or-flight, and cold water produces an initial fight-or-flight shock, then a sharp vagal rebound. Sauna has a better long-term case than cold exposure.
Which devices measure it properly
The best comparison so far tested five wearables overnight against a chest-strap heart tracing, in only 13 healthy adults over 536 nights — like all the device studies below, small and mostly in young, healthy people:[45]
| Device | Worn on | Typical per-night error |
|---|---|---|
| Oura Ring 4 | Finger | ~6% |
| Oura Ring 3 | Finger | ~7% |
| WHOOP 4.0 | Wrist | ~8% |
| Garmin Fenix 6 | Wrist | ~11% |
| Polar Grit X Pro | Wrist | ~16%, swinging by up to ±24% |
The authors rated Garmin's and Polar's agreement as poor. The Apple Watch was not in this study; tested separately, its HRV reading — which is SDNN, not RMSSD — ran about 8 ms low with a typical error near 29%, even though its heart rate was accurate.[46]
Don't read the table as "rings beat watches". The authors put Oura's lead down to processing, not finger placement. In another lab study of 53 people, the wrist-worn WHOOP did best and an older Oura did no better than the Apple Watch — though that lab gets research support from WHOOP.[47] Oura itself was accurate only when noisy stretches were filtered out and readings averaged over at least 30 minutes, and more than half of participants aged 45 and over had errors above 10%.[48] Whole-night averaging and good filtering seem to matter as much as where the sensor sits.
Daytime wrist readings are mostly noise from movement, light and extra beats. For a spot check, or for clinical-grade accuracy, use a chest strap such as the Polar H10 with a phone app like Kubios, lying down for 5 minutes and then standing for 3. Against laboratory equipment in 43 healthy adults aged 18–39, it gave effectively identical RMSSD.[49] It costs less than most fitness trackers.
Reading your own numbers
- Your baseline is the only comparison. HRV depends on age, sex, fitness, genes and what you did yesterday. A 30 ms RMSSD might be excellent for one 65-year-old and worrying for a 20-year-old endurance athlete. Give a new device 2–3 weeks to establish a baseline, then think in 7–14-day trends.
- Single mornings are noise. Caffeine, last night's drink, body position, breathing and hydration all shift the reading.
- Higher is not always better. Irregular rhythms — atrial fibrillation, frequent extra beats, sinus-node disease — produce falsely high HRV, and a wearable won't necessarily flag it. Heavy training can also push HRV up: in a study of 21 trained male triathletes, three weeks of overload raised the morning RMSSD of the 13 who did it while their running performance fell about 9%, and both reversed after a lighter taper week.[50]
- Medicines distort it. Beta-blockers, antidepressants, anticholinergics and several heart drugs change HRV directly. Don't read meaning into a change after starting a new medicine without your clinician.
- A drop doesn't tell you why. It could be a hard workout, an oncoming cold, alcohol, sleep apnoea, depression or an inflammatory flare. If your trend stays clearly down for more than a couple of weeks, get checked rather than trying to fix the number.
Further reading
- Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation 1996 — the field's defining consensus.[51]
- Tsuji H, et al. Reduced heart rate variability and mortality risk in an elderly cohort. The Framingham Heart Study. Circulation 1994.[52]
- Jarczok MN, et al. Heart rate variability in the prediction of mortality: A systematic review and meta-analysis of healthy and patient populations. Neurosci Biobehav Rev 2022 — about 38,000 participants.[53]
- Umetani K, et al. Twenty-four hour time domain heart rate variability and heart rate: relations to age and gender over nine decades. J Am Coll Cardiol 1998.[54]
- Schumann A, et al. Quantifying cardiovascular autonomic aging with machine learning. Am J Physiol Heart Circ Physiol 2025.[55]
- Banerjee A. Understanding the shortcomings of heart rate variability as a tool for autonomic analysis. Front Physiol 2026.[56]
- Manresa-Rocamora A, et al. Heart rate variability-guided training for enhancing cardiac-vagal modulation, aerobic fitness, and endurance performance: a methodological systematic review with meta-analysis. Int J Environ Res Public Health 2021.[57]
- Laborde S, et al. Effects of voluntary slow breathing on heart rate and heart rate variability: A systematic review and a meta-analysis. Neurosci Biobehav Rev 2022.[58]
- Brown L, et al. The effects of mindfulness and meditation on vagally mediated heart rate variability: a meta-analysis. Psychosom Med 2021 — no significant effect on resting vagal HRV.[59]
- de Zambotti M, et al. Impact of evening alcohol consumption on nocturnal autonomic and cardiovascular function in adult men and women: a dose-response laboratory investigation. Sleep 2021.[60]
- Dial MB, et al. Validation of nocturnal resting heart rate and heart rate variability in consumer wearables. Physiol Rep 2025 — 13 adults, 536 nights.[61]
- Cooper TM, et al. Heart rate variability predicts levels of inflammatory markers: evidence for the vagal anti-inflammatory pathway. Brain Behav Immun 2015.[62]
- Van Assche E, Schiweck C. Resting state heart rate variability in depression: an introductory narrative review of cross-sectional and longitudinal evidence. J Pers Med 2026.[63]