HRV Normal Range by Age: What a Good Score Is and How to Improve It

A single HRV reading says little because HRV swings a lot from day to day and depends on how it was measured. The trend over weeks is what counts.

Your heart does not beat like a metronome. Between two beats there are sometimes 890, sometimes 940, sometimes 910 milliseconds. That fluctuation is heart rate variability, HRV for short. It arises because your autonomic nervous system constantly co-regulates the heart: the vagus nerve brakes, the sympathetic system accelerates. The finer this back-and-forth, the larger the fluctuation. That is why a high resting HRV is read as a sign of a well-recovered system and a low HRV as a sign of load, whether from stress, infection, alcohol or a hard session the day before.

Three metrics keep coming up, and they do not measure the same thing:

• RMSSD: the short-term beat-to-beat fluctuation, in milliseconds. It mostly reflects the vagus and is the number Oura, Whoop and Garmin display as HRV.
• SDNN: the overall spread of all beat intervals in the recording window. It grows with recording length, so a 24-hour SDNN is by definition larger than a 5-minute SDNN. The Apple Watch shows SDNN.
• pNN50: the share of consecutive beats that differ by more than 50 ms. Crude but intuitive.

In 1996 a joint Task Force of the European and North American cardiology societies set the measurement standards ^[5]. Its central point: 5-minute short-term values and 24-hour long-term values cannot be compared, and every value needs its conditions stated (lying, sitting, time of day, breathing).

On top of that, HRV swings strongly from day to day, in healthy people just as in professional athletes. Sports scientists who monitor endurance athletes therefore advise never to judge a single reading. They compare rolling weekly averages with the personal trend instead ^[11]. Psychological stress also lowers HRV: a 2018 review of 37 studies, mostly with short lab or exam stress, found lower vagal activity under stress in most of them ^[10]. A single value on a stressful Tuesday tells you little. The trend over weeks tells you a lot.

The most reliable chart for Germany comes from the KORA S4 study in the Augsburg region. Researchers led by Andreas Voss analysed the HRV of 1,906 healthy adults aged 25 to 74 in 2015, after excluding everyone with heart disease, diabetes, hypertension or relevant medication ^[1]. The measurement: 5 minutes of ECG lying down after 5 to 10 minutes of rest. That is the classic lab value. The values below are rounded means; the study reports pNN50 as a fraction (for example 0.23), here it is given in percent.

Age	RMSSD women	RMSSD men	SDNN women	SDNN men	pNN50 women	pNN50 men
25 to 34	43 ms	40 ms	49 ms	50 ms	23 %	20 %
35 to 44	35 ms	32 ms	45 ms	45 ms	16 %	13 %
45 to 54	26 ms	23 ms	37 ms	37 ms	8 %	6 %
55 to 64	21 ms	20 ms	31 ms	33 ms	5 %	4 %
65 to 74	19 ms	19 ms	28 ms	30 ms	4 %	4 %

Source: Voss et al. 2015, 5-minute ECG lying down, 782 women and 1,124 men ^[1].

Three things stand out. First, RMSSD falls to about half between the mid-20s and the mid-60s and flattens after that. Second, the spread is enormous: in women aged 25 to 34, RMSSD was 42.9 ms plus or minus 22.8 ms, so the standard deviation is about half the mean. A value of 25 ms is entirely normal in this group, and so is 65 ms. Third, within the age groups there was no statistically meaningful difference between women and men for RMSSD, SDNN and pNN50 ^[1]. The separate columns mainly show you how small the sex difference is.

For comparison, a 2010 systematic review pooled 44 studies with 21,438 healthy adults ^[2]. For 5-minute recordings it found a mean RMSSD of 42 ms (from the 15 studies that reported it; study means ranged from 19 to 75 ms) and a mean SDNN of 50 ms (from 27 studies; 32 to 93 ms). The Augsburg values therefore sit within the international range, toward the lower end because the sample contains many older adults.

Want to know where your own number falls, not just the average? A Dutch study of 84,772 healthy people aged 13 to 91 from the Lifelines cohort reports percentiles ^[12]. It used 10-second resting ECGs, a much shorter window. Its medians land close to the Augsburg means, but the spread is wider. Read the range like this: 96 out of 100 healthy people your age sat between the two numbers.

Age	Women: median (2nd to 98th percentile)	Men: median (2nd to 98th percentile)
20 to 24	52 ms (11 to 206)	48 ms (10 to 174)
25 to 29	48 ms (12 to 181)	42 ms (10 to 173)
30 to 34	42 ms (12 to 161)	37 ms (10 to 141)
35 to 39	38 ms (11 to 142)	33 ms (9 to 124)
40 to 44	34 ms (10 to 124)	29 ms (8 to 106)
45 to 49	29 ms (8 to 110)	26 ms (7 to 96)
50 to 54	27 ms (8 to 96)	24 ms (7 to 88)
55 to 59	23 ms (7 to 84)	21 ms (6 to 90)
60 to 64	21 ms (6 to 79)	19 ms (5 to 87)
65 to 69	18 ms (5 to 83)	18 ms (5 to 111)
70 to 74	18 ms (5 to 116)	16 ms (5 to 162)
75+	16 ms (4 to 106)	15 ms (4 to 130)

Source: Tegegne et al. 2020, Lifelines cohort, RMSSD from a 10-second resting ECG. The study excluded anyone with heart disease, high blood pressure, type 2 diabetes or obesity, and anyone taking antidepressants, beta blockers or other drugs that act on the vagus ^[12].

Two things differ from Augsburg. Here, women aged 20 to 45 had a median about 5 ms higher than men. That gap is small but statistically significant. It shrank after 45 and was gone after 60 ^[12]. And the lower limit barely moves with age: from 20 on, the 2nd percentile stays between about 4 and 12 ms.

A different scale applies to 24-hour recordings. Umetani and colleagues analysed long-term ECGs of 260 healthy people aged 10 to 99 in 1998 ^[3]. RMSSD and pNN50 fell fastest. From the sixth decade of life (your 50s) they levelled off at 47 and 24 percent of their starting value. SDNN declined more slowly, to 60 percent in the tenth decade (your 90s). Under 30, women had lower values than men; from 50 the difference disappeared. The absolute 24-hour numbers are much higher than the 5-minute values above and do not belong in the same chart.

A good HRV score is one that sits above your own average, measured with the same device at the same time. The chart above tells you where people your age land in the lab. It does not tell you whether you are healthy, and it says nothing about your wearable value, because that is produced under different conditions.

Still, many people ask for concrete numbers. Here is how to read them, separated by measurement type:

HRV of 20. As a 5-minute lab value that is about average from 55 on ^[1] and roughly the median at 60 to 64 in the Dutch data ^[12]. In a 30-year-old it is below the median, but still above the lowest 2 percent, which sit around 10 to 12 ms ^[12]. Low, not alarming.

HRV of 30. As a 5-minute RMSSD lying down, that is average from the mid-40s and below average in the late 20s, but within one standard deviation ^[1]. As a night-time wearable value, a 30 in a 30-year-old is below what most users see; in a 60-year-old it is unremarkable. In both cases the number alone is not a finding.

HRV of 40. That is about the lab mean at 25 to 34 and above the mean from 35 on ^[1]. In the Dutch data it is close to the median for women aged 35 to 39 and men aged 30 to 34, and above the median for everyone from 40 on ^[12]. As a night-time wearable value, judge it only against your own baseline.

HRV of 50. As a 5-minute RMSSD, 50 ms is above the mean in every age group ^[1]. If your number is an SDNN, 50 ms is about average for 25 to 34 ^[1]. As a night-time RMSSD from Oura, Whoop or Garmin, a 50 is not unusual between 30 and 50 years of age, but devices differ too much for a firm cutoff.

HRV of 100. In the 5-minute lab that is rare, more than twice the highest group mean of 43 ms ^[1]. At night on a wearable, young endurance-trained people can see such values. 100 is not a target, and anyone who does not reach it has done nothing wrong.

You will also find thresholds online such as SDNN under 50 ms unhealthy, 50 to 100 ms compromised, over 100 ms healthy. That scale comes from 24-hour risk assessment in cardiac patients and is classified exactly that way in review papers ^[4]. You must not apply it to a 5-minute value or to your watch. Mean 5-minute SDNN sits below 50 ms in almost every age group of the chart, so by that scale most healthy adults would count as unhealthy ^[1].

Two things help more than any threshold. First, the trend: compare your value with the rolling 7-day mean, as is standard in training monitoring ^[11]. A 7-day average that stays clearly below your usual range for several days is a signal; a single low morning is not. Second, context: stress lowers HRV ^[10], alcohol lowers it ^[9], an infection lowers it. If you know the reason, you do not need the chart.

The most common mistake in HRV comparisons: users put their night-time value next to a lab chart and get alarmed or pleased for no reason. Wearables measure at night, over hours, from the pulse signal at the wrist or finger. The chart above was recorded during the day, for 5 minutes, by ECG. At night the vagus is more active, so night values usually come out higher. On top of that, every manufacturer calculates differently. This is how it stands today, according to the manufacturers:

• Oura shows the average night-time RMSSD in milliseconds, calculated from 5-minute windows across the whole night. In a small 2021 validation study (5 young adults, measured sitting and awake, not overnight) against a multi-lead ECG, the Oura RMSSD was off by 6.8 percent on average, with an agreement score of 0.91, where 1.0 would be a perfect match ^[6].
• Whoop also shows RMSSD, according to the manufacturer from the last deep-sleep phase of the night. That often yields higher and more volatile numbers than Oura's night average.
• Garmin calls it HRV Status: a night-time RMSSD-based average from 5-minute windows, shown as a night value and as a 7-day mean, plus a personal normal range the watch builds from about three weeks of data. Balanced, Unbalanced and Low compare your 7-day average with your own baseline, not with a population chart. That is exactly how HRV should be read. Only the Poor label, which appears when your values average well below the normal range for your age, uses an age reference.
• Apple Watch has long shown SDNN from short samples taken during the day, among others during mindfulness sessions. SDNN is a different metric from RMSSD, the windows are short and the time of day varies. With Series 12 and Ultra 4 (September 2026), Apple added a Recovery HRV that is sampled whenever you are still, and the Vitals app now shows overnight HRV. Apple does not say whether the new value is SDNN or RMSSD. Either way, compare Apple values only with earlier Apple values, not with Oura or the chart.
• Chest strap plus app (Polar H10 with HRV4Training, Elite HRV or Kubios) gives a short morning reading lying or sitting, usually 1 to 5 minutes. That comes closest to the lab method. In the same small validation study, HRV4Training had a 4.1 percent error for RMSSD with an agreement of 0.98, while a pure phone-camera app had a 112 percent error ^[6].

Is there a chart for night-time values? The largest one comes from Fitbit. Its researchers analysed a randomly chosen 24-hour period of heartbeat data from 8,203,261 users in 5-minute windows. They published benchmark tables by age and sex for two times of day ^[13]. The pattern matches the lab: RMSSD falls with age, women and men barely differ on RMSSD, and HRV swings strongly across the day. People who walked more steps per day had higher HRV, and the more steps, the higher. Two caveats: Fitbit funded the study, and other devices calculate differently. Treat these tables as a rough frame, not a verdict on your Oura or Garmin number ^[13].

Three simple rules follow. First: never compare values from two devices with each other. Second: never compare night-time wearable values with the lab chart, only with your own history. Third: if you want a value that can be compared with studies, measure in the morning lying down with a chest strap, 5 minutes, always the same way. Which device suits which purpose is covered in detail in the HRV wearable comparison.

HRV can be influenced, but less dramatically than wearable marketing suggests. Here are the measures with the best data, sorted by evidence.

Slow breathing, around 6 breaths per minute. This is the best-studied single measure. A 2022 meta-analysis evaluated 223 studies on voluntarily slowed breathing and separated three time windows: during the breathing, right after a session, and after multi-week programmes ^[7]. Vagally mediated HRV rose in all three. The largest effect occurs during the breathing itself, because breathing rate directly drives heart rate. The carry-over to resting HRV after a multi-week programme is smaller, but present. In the studies people usually practised 5 to 20 minutes a day, often with about 5 seconds in and 5 seconds out. Costs nothing, has hardly any side effects.

Endurance training. A 2005 meta-analysis pooled controlled training studies. Across 13 studies with 322 people, the high-frequency HRV component (the vagal share) rose with an effect size of 0.48. Effect size is a standard yardstick for how big a change is, and around 0.5 counts as medium. Across 12 studies the resting beat interval rose with an effect size of 0.75, which means a slower resting pulse ^[8]. These are medium effects that take weeks to months. Older adults responded more weakly on resting heart rate than younger ones. What counts is regular base training, not intensity at any price. What that looks like is in the Zone 2 and VO2max guide; the VO2max calculator estimates your starting point.

Less alcohol. Probably the fastest lever. A Finnish study followed 4,098 employees with HRV measured during sleep ^[9]. It tracked HRV-based recovery in the first three hours of sleep (a score calculated from your heart rhythm) and saw it fall with every step up in dose:

• low dose (up to 0.25 g of alcohol per kg of body weight, a bit under half a litre of beer at 75 kg): down 9.3 percentage points
• moderate dose (0.25 to 0.75 g/kg): down 24 percentage points
• high dose (over 0.75 g/kg): down 39.2 percentage points

The pattern held in women and men alike. Anyone who checks their night value the morning after two glasses of wine sees this directly.

Sleep and regularity. Short, late or irregular sleep lowers the night value for most users, and wearables make that visible. Clean effect sizes from large studies are missing here, though; the evidence is consistent rather than precise. What is proven for sleep duration and quality is in the sleep guide.

Stress and weight. Psychological stress lowers HRV. A 2018 review of 37 studies, mostly with short lab or exam stress, found lower vagal activity under stress in most of them ^[10]. For long-term stress the data are thinner. What helps against it is individual; the breathing exercise above is one of the few measures with numbers. For weight loss there are hints of rising HRV in severe obesity, but no effect size we could quote here in good conscience.

What you can skip. For supplements, cold plunges or red light there are no solid data showing that they raise resting HRV for good. A cold plunge changes the value the same evening, nothing more.

HRV is not a diagnostic value. No doctor diagnoses a disease from your watch, and no number in the chart above marks the line between healthy and sick. The chart comes from people who were screened as healthy beforehand, and even so some of them sat far below the mean ^[1]. Umetani and colleagues showed this in the over-65s: 25 percent of healthy older adults were below the published risk thresholds for the SDNN index, 12 percent for RMSSD, without any disease ^[3]. Thresholds from patient studies simply do not fit healthy people.

What usually sits behind a low value, in this order: an incoming infection, alcohol the evening before, little or late sleep, a hard training session, acute stress, travel across time zones, heat, a late large meal. In women HRV also varies across the cycle, usually a little lower in the second half ^[15]. All of it is expected and passes.

Measurement errors are common too. Extrasystoles, meaning single extra beats, push RMSSD up because the algorithm reads them as a huge fluctuation. A suddenly very high value is therefore no more a success than a low one is an alarm. With atrial fibrillation, the HRV calculation stops making sense altogether ^[14].

When to see a doctor is decided by symptoms, not by the number:

• palpitations, a racing heart or an irregular pulse that you feel yourself or that the watch flags as possible atrial fibrillation
• dizziness, fainting, chest pain or shortness of breath, especially on exertion
• a resting heart rate that stays clearly higher than usual for weeks without an obvious reason
• an HRV drop over two to three weeks together with exhaustion, sleep problems or weight loss

In these cases an ECG or long-term ECG at your GP or cardiologist is the right step, and it is because of the symptoms. The HRV trend from your watch can be a useful conversation starter. Without symptoms, a low value is mainly a prompt to look at sleep, alcohol and load. And if you stay below your baseline for longer without finding a reason, a check-up is a sensible decision, not an obligation.