Heart Rate Variability (HRV) Explained: What It Is & How to Improve It

A good HRV is whatever is normal for you. Across healthy adults, average RMSSD sits near 40 milliseconds in the late twenties and falls to around 19 ms by the late sixties, but the spread inside each age bracket is so wide that the population average tells you little about your own reading.

Almost every HRV chart online gives numbers with no source attached. Below are reference values with the study, the sample size and the measurement conditions behind them.

Normal HRV by age and sex

These values come from Voss and colleagues, published in PLOS ONE in 2015: five minute ECGs in 1,906 healthy adults, 782 women and 1,124 men, aged 25 to 74, all recorded lying down after five to ten minutes of rest. Values are mean plus or minus one standard deviation, in milliseconds.

Women, short term HRV by age

Men, short term HRV by age

Source: Voss A, Schroeder R, Heitmann A, Peters A, Perz S. Short-Term Heart Rate Variability, Influence of Gender and Age in Healthy Subjects. PLOS ONE 2015;10(3):e0118308. Five minute supine ECG recordings. No data below age 25 or above age 74.

Read the standard deviations before you read the means. A woman aged 35 to 44 has an average RMSSD of 35.4 ms, but the standard deviation is 18.5 ms. That means roughly two thirds of healthy women her age fall somewhere between 17 and 54 ms. The table describes a population. It does not describe you.

A larger pooled dataset agrees on the scale of the spread. Nunan, Sandercock and Brodie pooled 44 studies covering 21,438 healthy adults and reported a mean SDNN of 50 ms (standard deviation 16) and a mean RMSSD of 42 ms (standard deviation 15). Across those studies, reported SDNN ranged from 32 to 93 ms and RMSSD from 19 to 75 ms, all in healthy people.

What your wearable measures, and when

Before you compare your number to the table, check what your device actually calculates. They do not all measure the same thing, or at the same time.

Sources: Apple Developer documentation for HealthKit heartRateVariabilitySDNN; Oura Member Care, Heart Rate Variability; Garmin, HRV Status; WHOOP, Heart Rate Variability: How to Read and Improve Your HRV; and Nasr A, Behbehani K, Wang J, Pierson C, Jarrett RB, Greer T, Zafereo J, Digital Health 2025;11:20552076251394960, which documents Fitbit's method.

Apple is the outlier. HealthKit stores HRV as SDNN, a broader measure than RMSSD. An Apple Watch reading of 45 and a Garmin reading of 45 are not the same thing, and neither is comparable to a five minute supine RMSSD from the table above. Pick one device and compare it only to itself.

What HRV actually reflects

Your heart does not beat like a metronome. At 60 beats per minute, the gaps between beats are not all exactly one second. HRV is the size of that variation, measured in milliseconds.

Most of that beat to beat wobble comes from your vagus nerve, the main pathway of the parasympathetic nervous system. Vagal signalling is fast enough to nudge the timing of the very next beat. Sympathetic signalling, the fight or flight side, works through slower chemical pathways and cannot change things beat by beat. That difference in speed is why RMSSD, which looks only at the gap between consecutive beats, is treated as a marker of vagal activity.

SDNN is the standard deviation of every interval in a recording. It captures the fast vagal variation plus slower rhythms, including breathing and blood pressure regulation. A broader signal, and a less specific one.

Be clear about what HRV is not. It is not a fitness test and it is not a diagnosis. It measures autonomic state, which responds to training load, illness, alcohol, poor sleep, heat and a difficult day at work without distinguishing between them. A low reading tells you your nervous system is under load. It does not tell you why.

Why comparing your HRV to someone else's is close to useless

The between person variation in HRV is enormous. Nunan and colleagues noted that in their review some measures differed by as much as 260,000 per cent between individuals within the same study, particularly in the frequency domain measures. Even the more stable time domain measures, as the tables above show, carry standard deviations close to half the mean.

Some of that is genetic, some is age, some is sex. None of it is a target. A 55 year old man with an RMSSD of 19 ms sits exactly on the average for his age. A 30 year old woman with the same reading is well below hers. Neither fact is actionable on its own.

What is actionable is direction. Your own HRV, measured the same way night after night and tracked over weeks, is a real signal. Your HRV against a stranger's is noise.

What actually raises HRV

Aerobic training

A 2024 meta-analysis of randomised controlled trials in healthy adults by Amekran and El Hangouche pooled 16 trials covering 623 participants. Exercise training improved RMSSD with a standardised mean difference of 0.84 (95% CI 0.36 to 1.31), SDNN by 0.58 (0.16 to 1.00) and high frequency power by 0.89 (0.27 to 1.51).

That is a moderate to large effect, but be honest about the evidence base: 16 trials and 623 people is small, and the confidence intervals are wide. The direction is consistent, the precise size is not settled.

Training also raises HRV over months and lowers it over hours. A hard session on Tuesday usually shows up as a lower reading on Wednesday morning. That drop is the training working, not failing.

Alcohol

This is the most reproducible finding in the practical literature, and it holds in the lab and in the wild.

The largest real world dataset comes from Grosicki and colleagues, published in PLOS Digital Health in 2026, covering 20,968 people and 5,109,185 person days of wearable data. Comparing each person against their own drinking history, one drink more than their personal average, versus one less, was associated with nocturnal HRV about 3.8 ms lower in females (99.9% CI 4.1 to 3.5) and 3.3 ms lower in males (3.5 to 3.1). Nocturnal resting heart rate rose by 2.8 and 2.4 beats per minute respectively. Effects were larger in women and in younger adults.

A separate real world sample of 4,098 Finnish employees, across 12,411 recording days, measured the acute effect during the first three hours of sleep. RMSSD fell by 2.0 ms after a low dose (up to 0.25 g of alcohol per kilogram of body weight), 5.7 ms after a moderate dose (0.25 to 0.75 g/kg) and 12.9 ms after a high dose (above 0.75 g/kg). For a 70 kg adult, that high threshold works out at roughly five standard Australian drinks. On alcohol free nights the researchers saw recovery build as sleep progressed. After a high dose, it did not.

Both of those are observational. The causal evidence is smaller and points the same way: a single blind, placebo controlled laboratory study of 26 healthy adults aged 30 to 60 found both RMSSD and SDNN reduced after low and high alcohol doses compared with a dealcoholised placebo, with the high dose producing the larger effect.

Alcohol is the fastest way to move your HRV, and it moves it down. The Grosicki data also found that drinking earlier in the day, sleeping longer afterwards and keeping the next day lighter each reduced the disruption.

Slow paced breathing

Laborde and colleagues screened 1,842 abstracts and included 223 studies in a 2022 systematic review and meta-analysis. Vagally mediated HRV increased during voluntary slow breathing (172 studies), immediately after a single session (16 studies) and after a multi-session intervention (49 studies).

Two caveats. The rise during the breathing itself is partly mechanical, because breathing slowly directly enlarges the beat to beat swing. The more interesting result is the carryover after the session ends. And the review measured HRV, not health outcomes.

Sleep

Sleep is where the evidence is thinner than the popular claims suggest. The alcohol data does show that longer sleep after drinking reduced the HRV disruption, which is consistent with sleep being protective. That is not the same as proving an extra hour of sleep raises a healthy person's HRV. Treat a rise in HRV as a plausible side effect of sleeping well, not the goal.

Why the measurement conditions change the number

Three things change your reading without anything changing in your physiology: when you measure, what position you are in, and how long the recording runs.

The reference table above was built from five minute recordings taken lying down after a rest period, because posture and recent activity move the number substantially. A reading taken sitting at your desk is not comparable to that table, and not comparable to one you took standing up last Tuesday.

Time of day matters for the same reason. Coffee, stairs, a tense meeting and lunch each shift autonomic state for a while afterwards. That is why the wearables that take HRV seriously measure during sleep, when the confounders are largely absent and every night is measured under roughly the same conditions.

Night and morning readings are related but not interchangeable. In a three week study of 11 young endurance athletes, nocturnal and morning RMSSD correlated strongly, r = 0.895 in week one and r = 0.878 in week three, but the absolute values differed significantly between methods and the morning readings varied more day to day. That is a small sample of young athletes, so treat it as indicative rather than settled. Both work if you are consistent. Neither works if you switch between them.

What a useful week actually looks like

Take a 38 year old woman whose seven day average nightly RMSSD sits at 34 ms. Against the reference table she is close to the middle of her age group, where the average is 35.4 ms. Context, not a score.

Her week, measured on the same device every night:

• Monday 36 ms. Rest day.

• Tuesday 33 ms. Intervals on Tuesday evening. A small dip the following night is expected.

• Wednesday 22 ms. Three glasses of wine at a work dinner, finished late.

• Thursday 29 ms. Still below her baseline, still recovering.

• Friday 35 ms. Back to normal.

• Saturday 31 ms. Long run Saturday morning.

• Sunday 37 ms.

Her seven day average is 31.9 ms against a baseline of 34. That is a mild dip, fully explained by one heavy night and two hard sessions, and it needs no intervention.

Notice what the single numbers are worth. Wednesday's 22 ms would sit near the bottom of her age group in the reference table, and it means nothing about her health. It means she drank three glasses of wine. Her 12 ms drop is larger than the population estimate of roughly 3 to 4 ms per drink would predict, which is what you should expect: population averages describe groups, not individuals.

The pattern that would warrant a change is different in shape. If her seven day average drifted from 34 to 28 to 25 across three weeks with no alcohol and no unusual training to explain it, that is a trend. The sensible response is to reduce load and protect sleep until it recovers. If it stays suppressed, that is a conversation with a GP.

Where your own data fits

This is the argument for reading HRV as a trend rather than a number, and it is how hlth. coach uses it. Your nightly reading sits against your own rolling baseline alongside training load and sleep, so a dip after a hard session reads as a hard session, and a dip that persists across weeks reads as something worth acting on.

The short version

• There is no universal good HRV. Reference values show where a population sits, and their standard deviations are close to half the mean.

• In healthy adults aged 25 to 34, average RMSSD is about 43 ms in women and 40 ms in men. By 65 to 74 it is about 19 ms in both.

• HRV mostly reflects vagal, parasympathetic activity. It is not a fitness test and not a diagnosis.

• Between person differences are so large that comparing your number to anyone else's is close to meaningless.

• Devices are not interchangeable. Apple Watch reports SDNN, most others report RMSSD, and they sample at different times.

• Aerobic training raises HRV over months and lowers it for a day or two after a hard session.

• Alcohol is the most reliable suppressor, and the effect scales with dose.

• Slow paced breathing raises vagally mediated HRV during, immediately after, and following repeated practice.

• Measure the same way every time, ideally overnight, and watch the rolling trend across weeks rather than single nights.

Common questions

What is a good HRV?

A good HRV is one that sits in a normal range for your age and stays steady relative to your own baseline. Population ranges are wide and overlapping, so a number that looks low against someone else's can be perfectly normal for you. Your own trend matters more than any published average.

What is a normal HRV by age?

HRV falls with age, and the spread within any age group is wide. The normative tables above give published ranges by age and sex. Use them to check you are somewhere in the expected region, not to set a target.

Why is my HRV different on my watch than my friend's?

Because devices measure different things in different ways. Some use RMSSD, others report a proprietary score. Some sample across the whole night, others take a short window. Two devices on the same wrist can disagree substantially, so comparing across brands tells you very little.

Does a low HRV mean something is wrong?

Not on its own. HRV drops in response to hard training, poor sleep, alcohol, illness and stress, and single readings are noisy. A one day dip is rarely worth acting on. A sustained fall over a week or more, alongside how you actually feel, is the signal worth reading.

How can I improve my HRV?

The reliable levers are unglamorous: consistent sleep, sensible training load, less alcohol, and aerobic base work. Interventions marketed specifically at raising HRV are on much thinner evidence than the basics.

Should I train if my HRV is down?

Usually yes, but adjusted. A single low reading is not a reason to stop, and treating it as one leads to skipped sessions on noise. A run of low readings, combined with poor sleep and heavy legs, is a reason to move the hard session and do something easier.

Sources

• Voss A, Schroeder R, Heitmann A, Peters A, Perz S. Short-Term Heart Rate Variability, Influence of Gender and Age in Healthy Subjects. PLOS ONE 2015;10(3):e0118308. doi:10.1371/journal.pone.0118308

• Nunan D, Sandercock GRH, Brodie DA. A Quantitative Systematic Review of Normal Values for Short-Term Heart Rate Variability in Healthy Adults. Pacing and Clinical Electrophysiology 2010;33:1407 to 1417. doi:10.1111/j.1540-8159.2010.02841.x

• Amekran Y, El Hangouche AJ. Effects of Exercise Training on Heart Rate Variability in Healthy Adults: A Systematic Review and Meta-analysis of Randomized Controlled Trials. Cureus 2024;16(6):e62465. doi:10.7759/cureus.62465

• Grosicki GJ, Robinson AT, Joyner MJ, Carter JR, von Hippel W, Presby DM, et al. Real-world effects of alcohol on heart rate, sleep, and physical activity by age and sex. PLOS Digital Health 2026;5(3):e0001284. doi:10.1371/journal.pdig.0001284

• Pietilä J, Helander E, Korhonen I, Myllymäki T, Kujala UM, Lindholm H. Acute Effect of Alcohol Intake on Cardiovascular Autonomic Regulation During the First Hours of Sleep in a Large Real-World Sample of Finnish Employees: Observational Study. JMIR Mental Health 2018;5(1):e23. doi:10.2196/mental.9519

• de Zambotti M, Forouzanfar M, Javitz H, Goldstone A, Claudatos S, Alschuler V, Baker FC, Colrain IM. Impact of evening alcohol consumption on nocturnal autonomic and cardiovascular function in adult men and women: a dose-response laboratory investigation. Sleep 2021;44(1):zsaa135. doi:10.1093/sleep/zsaa135

• Laborde S, Allen MS, Borges U, Dosseville F, Hosang TJ, Iskra M, Mosley E, Salvotti C, Spolverato L, Zammit N, Javelle F. Effects of voluntary slow breathing on heart rate and heart rate variability: A systematic review and a meta-analysis. Neuroscience and Biobehavioral Reviews 2022;138:104711. doi:10.1016/j.neubiorev.2022.104711

• Mishica C, Kyröläinen H, Hynynen E, Nummela A, Holmberg HC, Linnamo V. Evaluation of nocturnal vs. morning measures of heart rate indices in young athletes. PLOS ONE 2022;17(1):e0262333. doi:10.1371/journal.pone.0262333

• Nasr A, Behbehani K, Wang J, Pierson C, Jarrett RB, Greer T, Zafereo J. Agreement between Fitbit Versa 4 and a reference standard on time-domain nocturnal heart rate variability: A pilot study. Digital Health 2025;11:20552076251394960. doi:10.1177/20552076251394960

• Device documentation: Apple Developer Documentation, HKQuantityTypeIdentifier heartRateVariabilitySDNN, developer.apple.com; Oura Member Care, Heart Rate Variability, support.ouraring.com; Garmin Technology, HRV Status, garmin.com; WHOOP, Heart rate variability: how to read and improve your HRV, whoop.com

This article is general information only. It is not medical advice, and it is not a substitute for assessment by a qualified health professional who knows your history.