Normal HRV by Age and Sex

Among 5,507 healthy Dutch men aged 40 to 44, the median resting RMSSD was 29.0 ms. The healthy range, from the 2nd to the 98th percentile, ran from 8.1 ms to 105.5 ms, a thirteenfold spread inside one five-year age band. The figures come from Tegegne and colleagues' 2020 reference values in the European Journal of Preventive Cardiology. A 42-year-old man who searches "is 30 ms HRV good" gets a clean answer from it. His number sits almost exactly in the middle.

That answer holds only for RMSSD taken from a 10-second resting ECG. If his 30 ms came from an Apple Watch, it is SDNN from a different kind of recording, and the Lifelines table says little about it. If it were a 24-hour SDNN from a Holter monitor, a cardiologist would read it as a warning. The same two digits carry three different meanings.

Below are the age and sex stratified data that exist, each with its metric and recording window named, followed by the case for comparing your number with your own recent history. The HRV and recovery readiness guide covers measurement protocols and training decisions. This piece stays on one question, what your number means.

01What the number measures

At 60 beats per minute the gap between beats might run 980 ms, then 1,040 ms, then 1,010 ms. Heart rate variability is a statistic computed on that series of gaps. At rest, most of the fast beat-to-beat variation comes from the vagus nerve, the parasympathetic brake on the heart, which lets heart rate rise a little on each inhale and pulls it back on each exhale. Hard training, short sleep, illness, alcohol and psychological stress all reduce vagal activity, and the intervals become more uniform.

The HRV glossary entry has the formal definitions. HRV is a summary statistic with several competing formulas, and every formula gives a different answer depending on how long the recording ran and what the body was doing.

02Why two devices never agree

Two friends comparing HRV are usually comparing different measurements that share a unit.

Different formulas

SDNN is the standard deviation of all normal beat-to-beat intervals in a recording. It captures total variability, fast and slow. RMSSD is the root mean square of the differences between successive intervals, which isolates the fast, mostly vagal component. Both are reported in milliseconds, on different scales. In Voss and colleagues' 2015 KORA sample in PLOS ONE, 5-minute recordings from women aged 25 to 34 produced a mean SDNN of 48.7 ms and a mean RMSSD of 42.9 ms. For women aged 55 to 64 the same two measures read 30.6 ms and 21.4 ms. The ratio between them shifts with age, so no fixed conversion factor turns one into the other.

Titan's own documentation states the same thing about its data. Apple Watch records HRV as SDNN, and Titan uses SDNN by default. Turning on Use RMSSD for HRV makes Titan compute RMSSD from the beat-to-beat Heartbeat Series in Apple Health instead. The Recovery settings article warns that "RMSSD and SDNN sit on different scales, so your old history no longer matches," and tells you to clear the Recovery cache after switching. If a metric switch breaks the comparison with your own last month, it breaks the comparison with a stranger too.

Different windows

SDNN grows with recording length, because longer recordings pick up slow oscillations from posture changes, temperature regulation and the day-night cycle. Shaffer and Ginsberg's 2017 review in Frontiers in Public Health is blunt about it: "it is inappropriate to compare metrics like SDNN when they are calculated from epochs of different length." In 3,387 adults, Munoz and colleagues (2015, PLOS ONE) found that a single 10-second RMSSD agreed with a 4 to 5 minute reference at r = 0.85 with almost no systematic bias. SDNN from the same 10 seconds agreed less closely and carried a large bias. RMSSD survives short windows far better than SDNN, which is why the big population studies report it.

The clinical SDNN bands that circulate online, under 50 ms unhealthy, 50 to 100 ms compromised and above 100 ms healthy, come from 24-hour ambulatory recordings (Shaffer and Ginsberg, 2017). A 24-hour SDNN includes a full day and night of heart rate swings. A watch reading covers a sliver of that, so holding a 35 ms watch reading against a 50 ms Holter cutoff tells you nothing about your heart.

Different timing and averaging

A reading during deep sleep and one taken at a desk after coffee catch the autonomic nervous system in different states. Devices also decide how to collapse many readings into one daily figure. Oura reports sleeping HRV. Apple Watch takes background readings a few times a day, at times you do not pick, as HRV not showing explains. Titan takes one daily value in order: HRV from a Mindfulness session started before noon, then the average of all readings during your sleep session, then the average of the past 24 hours (Recovery score). Two defensible rules applied to one heart produce two numbers.

03The population numbers

Each table below names its metric, window and sample. Numbers inside one table compare with each other. Numbers across tables mostly do not.

Median RMSSD by age and sex from Lifelines

Tegegne and colleagues (2020) built these reference values from the Lifelines Cohort in the northern Netherlands. After excluding people with cardiovascular disease, hypertension, diabetes, obesity or vagally active medication, and noisy recordings, they kept 84,772 participants aged 13 to 91. The metric is RMSSD from a standard 10-second resting ECG. The authors report the 2nd and 98th percentiles as the lower and upper limits of normal.

RMSSD in ms from a 10-second resting ECG. Source: Tegegne et al. (2020), Table 1.

The median falls by about 60 percent between the early twenties and the early sixties, then flattens. The lower limit of normal barely moves, from about 11 ms in young adults to about 5 ms after 60. The upper limit is enormous at every age. A 30-year-old woman at 15 ms and one at 150 ms both sit inside the healthy reference range.

Five-minute resting values from KORA

Voss and colleagues (2015) recorded 5 minutes of ECG, lying down after 5 to 10 minutes of rest, in 1,906 healthy adults aged 25 to 74 from the German KORA S4 cohort, 782 women and 1,124 men. They report means and standard deviations. RMSSD has a long right tail, so its means run higher than its medians.

Mean ± standard deviation in ms from a 5-minute supine recording. Source: Voss et al. (2015), Tables 5 and 7.

The 10-second Dutch data and the 5-minute German data land close together, as Munoz's result predicts. KORA women aged 45 to 54 averaged 26.3 ms. Lifelines women in the same decade had medians of 29.2 and 26.6 ms.

The other reference points people quote

The Nunan review itself warns that its studies disagreed partly because they cleaned beat data in different ways. The 24-hour bands apply only to 24-hour recordings. The Oura figures describe sleep, a different state from a daytime resting ECG.

04How HRV changes with age

Tegegne and colleagues' 2018 analysis of 149,205 Lifelines participants in Heart Rhythm states it in one line: "HRV strongly declined with age and was consistently higher in women." Age and sex together explained 17.4 percent of the variance in RMSSD. Physical activity, smoking, alcohol, stress, social well-being and neuroticism together added less than half a percent.

In the KORA data, mean RMSSD fell about 7.5 ms from the 25 to 34 decade to the next, another 9 ms by 45 to 54, then 3 to 5 ms more by 55 to 64. The step to 65 to 74 was not statistically significant in either sex. Lifelines shows the same bend, with the 25th percentile, median and 75th percentile falling steadily until about 60 and then holding. Umetani's 24-hour data from 1998 has the same pattern for RMSSD, which reached 47 percent of its value in the second decade of life by the sixth decade and then stabilized. SDNN over 24 hours declined far more slowly and still sat at 60 percent of its second-decade value in the tenth decade. The fast vagal component takes the biggest hit, mostly before 60.

Umetani also found something that should make anyone wary of clinical cutoffs. Among healthy subjects over 65, 12 percent had 24-hour RMSSD below published cutpoints for increased mortality risk, and 25 percent fell below the cutpoint on a related measure, the SDNN index. Healthy people read as high risk because the thresholds ignore age.

A comparison across a 20-year age gap, even with identical devices, is mostly a comparison of ages.

05Do women and men differ

A common claim online is that men have higher HRV. The studies disagree, in a pattern that shows once you sort them by metric.

Measures of total variability, SDNN and total spectral power, especially over 24 hours, lean male. Measures of the fast vagal component, RMSSD and high-frequency power, lean female or show no difference. Shaffer and Ginsberg's review notes lower SDNN in women "especially in 24 h studies." Every study here that followed the difference across the lifespan saw it shrink or vanish somewhere between 50 and 60.

The largest reliable sex gap in RMSSD is about 5 ms, in adults under 45. Age moves the Lifelines median by about 30 ms between the early twenties and the early sixties. Swapping SDNN for RMSSD moved the KORA values by 6 to 9 ms, more than the sex gap. If your HRV differs from your partner's, sex is the least likely explanation.

06What trained athletes see

Fitness offsets part of the age decline. Deus and colleagues (2019, Physiology and Behavior) recorded 10 minutes of seated HRV in master sprinters and master endurance athletes averaging 52 to 54 years old, untrained people of similar age, and untrained people around 25. The untrained middle-aged group averaged an RMSSD of 20.2 ms against 43.3 ms in the young controls. The master sprinters averaged 40.9 ms and the master endurance athletes 38.9 ms, and the age-related gap seen in untrained people did not appear in either athlete group. With 8 athletes per group, treat the size of the effect with caution. The direction matches Antelmi and colleagues (2004, American Journal of Cardiology). In 653 people without heart disease, HRV fell with age, differed by sex, and ran higher in those with greater functional capacity.

In Lifelines, physical activity explained a sliver of HRV variance, but that cohort measured everyday activity. The Deus athletes had trained for life, a dose few people in a general cohort reach.

In that sample a trained 52-year-old read like an untrained 25-year-old, so the age row matching an athlete's birthday describes someone else's body.

07Why your own baseline beats every chart on this page

A population table tells you where you sit among people of your age and sex measured the same way. It cannot tell you whether today is a good day. In Lifelines the healthy range for a 35-year-old man runs from 8.6 to 123.8 ms. Knowing you sit at the 40th percentile of that range says nothing about whether last night's 5 ms drop matters.

The within-person picture is different. Sinnreich and colleagues recorded 5-minute HRV twice, two months apart, in 70 adults and found correlations of 0.76 to 0.80 between sessions for high-frequency and total power. Their conclusion was that short HRV recordings are "stable over months and therefore characteristic of an individual." Between people HRV spreads across a tenfold range, while one person under a fixed protocol holds a stable level. The useful signal is the departure from that level.

Comparing with yourself holds age, sex, training history, device and metric constant. In Titan, your baseline is the median of your daily HRV and daily resting heart rate across your Recovery baseline window of 7, 30 or 60 days, with 60 as the default (baselines). A median ignores one strange night. Titan then measures today's distance from that median in median absolute deviations, your own typical day-to-day spread, rather than in milliseconds or percent. The Recovery score weights that HRV distance at +1.0 and the matching resting heart rate distance at −0.7, adds smaller terms for the change over the last 7 days and for how variable your HRV has been, and places the result against your last 60 Recovery calculations. Your 10th percentile maps to 0 and your 90th percentile maps to 100. The percentile chart survives, built from your own history in place of a stranger's.

Median absolute deviation fixes a problem that raw milliseconds and percentages both have. Take two cyclists on the same watch, both using SDNN. The first is 58, with a baseline median of 28 ms and a median absolute deviation of 3 ms. The second is 31, with a median of 71 ms and a median absolute deviation of 11 ms. Both wake up 8 ms below baseline. For the first rider that is 2.7 deviations down, far outside a normal morning. For the second it is 0.7 deviations, an ordinary Tuesday. A fixed millisecond rule treats them identically. A percentage rule, 29 percent against 11 percent, still ignores that some people's HRV swings widely every day and some barely moves.

The Vitals screen draws the same idea. For HRV it shows a dashed line at your median for the period you pick and shades the 20th to 80th percentile of your own daily values. That band is your personal norms table.

08Reading a low or high day

One low morning means little. Alcohol, a late meal or a loose watch band can pull a single reading down. When you want a rule of thumb outside the app, use this one: a morning HRV more than one median absolute deviation below your baseline median for two or more consecutive days, with resting heart rate above baseline, is a signal to pull intensity.

Each clause filters out a different false alarm. The deviation threshold scales to your own noise, so a drop that is routine for you does not trigger it. The two-day requirement screens out one-night disturbances. Resting heart rate rising at the same time adds a second signal pointing the same way, which makes a sensor or sleep-position artifact less likely. A pattern that keeps returning despite easier days can be an early sign of overreaching.

In Titan, a low day shows up as a Recovery score in the Low band, 0 to 39, with coaching text that suggests reducing intensity, prioritizing sleep and keeping to easy aerobic work. A score of 75 to 100 lands in the High band, which the app describes as a good day for quality training if it aligns with your plan. A high reading gives you room for a hard session. It does not require one.

09Your Recovery baseline window

The setting that decides what every comparison above is measured against is your Recovery baseline window. You choose it on the Recovery Baseline step during onboarding, from 7, 30 or 60 days. The default is 60, and it stays at 60 if you continue without changing it.

A short window follows recent change quickly, so a hard training block drags a 7-day baseline down within a week and a tired week can still produce decent Recovery scores. A 60-day window holds your longer normal and shows a rough week as a drop. For reading HRV against your own normal, 60 days is the better choice for most people.

To see which window you have, open You and look at the HRV Baseline and RHR Baseline tiles under Your Fitness, which carry a label such as "60 day baseline." The current version has no Settings screen for changing the window after onboarding. If you need a different one, reach out through the contact page. The baselines article lists which Titan score reads from which baseline.

10References

• Antelmi I et al. (2004). Influence of age, gender, body mass index, and functional capacity on heart rate variability in a cohort of subjects without heart disease. American Journal of Cardiology 93(3):381-385. https://doi.org/10.1016/j.amjcard.2003.09.065
• Deus LA et al. (2019). Heart rate variability in middle-aged sprint and endurance athletes. Physiology and Behavior 205:39-43. https://doi.org/10.1016/j.physbeh.2018.10.018
• Hughes L (2025, updated 2026). What is the average HRV? Oura blog. https://ouraring.com/blog/average-hrv/
• Kuo TBJ et al. (1999). Effect of aging on gender differences in neural control of heart rate. American Journal of Physiology, Heart and Circulatory Physiology 277(6):H2233-H2239. https://doi.org/10.1152/ajpheart.1999.277.6.H2233
• Munoz ML et al. (2015). Validity of (ultra-)short recordings for heart rate variability measurements. PLOS ONE 10(9):e0138921. https://doi.org/10.1371/journal.pone.0138921
• Nunan D et al. (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. https://doi.org/10.1111/j.1540-8159.2010.02841.x
• Shaffer F et al. (2017). An overview of heart rate variability metrics and norms. Frontiers in Public Health 5:258. https://doi.org/10.3389/fpubh.2017.00258
• Sinnreich R et al. (1998). Five minute recordings of heart rate variability for population studies, repeatability and age-sex characteristics. Heart 80(2):156-162. https://doi.org/10.1136/hrt.80.2.156
• Tegegne BS et al. (2018). Determinants of heart rate variability in the general population, the Lifelines Cohort Study. Heart Rhythm 15(10):1552-1558. https://doi.org/10.1016/j.hrthm.2018.05.006
• Tegegne BS et al. (2020). Reference values of heart rate variability from 10-second resting electrocardiograms, the Lifelines Cohort Study. European Journal of Preventive Cardiology 27(19):2191-2194. https://doi.org/10.1177/2047487319872567
• Umetani K et al. (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. https://doi.org/10.1016/S0735-1097(97)00554-8
• Voss A et al. (2015). Short-term heart rate variability, influence of gender and age in healthy subjects. PLOS ONE 10(3):e0118308. https://doi.org/10.1371/journal.pone.0118308