Treatment & Procedure

Heart rate variability (HRV)

Biohacking · RMSSD · SDNN

Your heart does not beat to the rhythm of a metronome. From the milliseconds between two beats, watch, ring and chest strap calculate a number. Which number that is is rarely stated — and the common metrics measure very different things.

In brief

RMSSD measures how much two adjacent heartbeats differ. Because only the vagus nerve can regulate that fast, this is predominantly the vagal brake. SDNN only becomes meaningful over 24 hours, pNN50 is a coarser variant of RMSSD, and the LF/HF ratio does not measure what its name promises. There are no normal values — only your own trend is meaningful.

Why only the fast fluctuation is vagal

The vagus nerve brakes the sinus node within a single heartbeat; the sympathetic nervous system accelerates more sluggishly, within seconds. Everything else follows from this: what fluctuates from beat to beat can almost only be vagally mediated.

This is exactly what RMSSD measures — the difference between two adjacent beats, squared and averaged. SDNN, by contrast, also captures everything slower: blood pressure regulation, thermoregulation, day-night rhythm. Over five minutes, it essentially measures the same as RMSSD, only less sharply.

Why the night value is the more useful one

HF band power rises at night anyway, the confounders of the day fall away, and a wearable measures every day under similar conditions during sleep. The third point is the most important — and it is pure standardization, not a better measurement.

The caveat remains: RMSSD reflects vagal modulation, not baseline tone. Changes in breathing rate and breathing depth shift RMSSD, pNN50 and HF power considerably without anything changing in vagal tone. Anyone with a cold gets a different value with the same recovery.

Six breaths per minute

At about six breaths per minute, breathing is at 0.1 hertz — where the blood pressure reflex also operates. The system resonates at its natural frequency, and heart rate variability rises visibly. In adults, however, the individual resonance frequency lies between 4.5 and 6.5 breaths per minute; six is the average, not the prescription.

What is well supported

In the authoritative review, RMSSD is expressly the primary time-domain metric for the vagally mediated components of heart rate variability; a vagal blockade practically eliminates the HF oscillations. The link with hard endpoints is also supported — in the Framingham study, LF power one standard deviation lower was associated with 1.70 times the risk of death. And slow breathing acutely raises vagally mediated heart rate variability, supported across 223 studies.

What the studies show

Framingham: 736 older people, four years

Two hours of ambulatory ECG from the original Framingham cohort were analyzed: 736 people, average age 72, 74 deaths in four years. One standard deviation less LF power corresponded to 1.70 times the risk of death (95 % confidence interval 1.37 to 2.09). In the younger cohort of 2,501 people without heart disease, one standard deviation of SDNN corresponded to a hazard ratio of 1.47 for cardiac events.

Normal values: 44 studies, 21,438 participants, no reference

The only systematic compilation found values consistently below the 1996 reference values and inter-individual variations of up to 260,000 percent. On top of that come opposing age trends: in 1,743 people between 40 and 100 years old, SDNN fell linearly, while rMSSD and pNN50 followed a U-shape, with the low point between 60 and 69 years.

Biofeedback: large effect on stress, small on depression

Across 24 studies with 484 participants, the effect on self-reported stress and anxiety was Hedges' g = 0.83 versus controls. For depressive symptoms, 14 randomized trials with 794 participants found g = 0.38, prediction interval −0.19 to 0.96. In cardiovascular patients, systolic blood pressure fell by 3.08 mmHg; depression and anxiety did not improve significantly.

Where the data stop

Every app scale that classifies your value as low or normal does so without a published reference. The Framingham finding is also not an argument for self-measurement: it was measured with two hours of ECG in people with an average age of 72, and the prediction was carried by LF power, not RMSSD.

The most widespread claim does not stand up to scrutiny: a drop over several days does not indicate overtraining. A meta-analysis of 24 studies found resting heart rate variability largely unaffected by overload; with declining performance, resting RMSSD even rose slightly. In 34 elite female wrestlers, swings in both directions marked the transition to non-functional overreaching. For an infection, by contrast, there is a real association — but the models use the daily pattern and other variables, and in 14 female athletes the breathing rate responded three days before the positive test, heart rate variability only one day before.

How to do it

Look at RMSSD, or its natural logarithm. You can ignore the LF/HF ratio: in the Framingham cohort with 2,501 participants, it was the only measure with no association with cardiac events.

Compare only with yourself. For the morning value, you need at least three valid measurements per week; for the nightly wearable value, at least five of seven nights. Your baseline range shifts over months: endurance training raised RMSSD by a standardized mean difference of 0.84 across 16 randomized trials.

For resonance breathing, start at six breaths per minute and also try five and seven; in the comparative study, participants breathed for 15 minutes at a stretch. Schedule a fixed time — in a remote study, only 16.5 percent of 127 employees managed half of the sessions.

Safety

The measurement is risk-free, and so is the breathing exercise. Anyone who notices dizziness or tingling is breathing too deeply and should breathe more shallowly instead of lowering the rate further.

The number is not a finding: a low value is not a diagnosis, a high one not a certificate of health. Anyone who notices an irregular pulse, palpitations, shortness of breath or chest pain should have it checked by a physician — an arrhythmia distorts every metric. HRV biofeedback has consistently been studied as a complement, not as a replacement for treatment.

BK-Score Supported, with caveats

Human evidence		6
Mechanism		6
Safety data		8
Hype gap		3
Track record of use		8

The measured variable itself is well studied: two Framingham cohorts with 736 and 2,501 people, a review of normal values covering 21,438 participants and biofeedback meta-analyses covering a total of 51 studies (24 on stress and anxiety, 14 randomized on depression, 13 randomized in cardiovascular patients). What is not supported is its use as a daily steering tool – the meta-analysis of HRV-guided versus pre-planned training found an advantage only in the vagal HRV values themselves, and no significant one in maximal oxygen uptake and endurance performance (Manresa-Rocamora 2021); usable normal values do not exist, and resting HRV does not carry the overtraining signal. The vagal side is well understood, the sympathetic side is not: there is no clean measure for it in a short-term ECG, and the widespread LF/HF ratio is considered outdated. What is advertised is decision validation; what exists is measurement validation.

The score rates the state of knowledge, not the effect. Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about HRV

Which metric should I look at?

RMSSD, or its natural logarithm. It reflects the vagal brake most directly and is the value research works with. SDNN only works over 24 hours, pNN50 measures the same thing, only more coarsely.

Why does my app show a stress level based on LF/HF?

Because that is a manufacturer convention. About half of the variability in the LF band itself comes from the parasympathetic nervous system, and with slow breathing, respiration moves into this band. In the Framingham cohort with 2,501 participants, LF/HF was the only measure with no association with cardiac events.

Is my value normal?

The question cannot be answered. The compilation of 44 studies with 21,438 healthy adults found inter-individual variations of up to 260,000 percent. Only your own trend is meaningful.

Does heart rate variability decline with age?

Not equally for all metrics. In 1,743 people between 40 and 100 years old, SDNN fell linearly, while rMSSD and pNN50 followed a U-shape, with the low point between 60 and 69 years and rising values again afterwards.

My value has been low for days — am I overtrained?

The drop does not support that conclusion. The meta-analysis of 24 studies found resting heart rate variability largely unaffected by overload, and with declining performance resting RMSSD even rose slightly. In 34 elite female wrestlers, swings in both directions indicated the transition to non-functional overreaching.

How much does breathing at six breaths per minute bring?

Vagally mediated heart rate variability rises acutely, supported across 223 studies. For complaints, the figures are mixed: Hedges' g = 0.83 for stress and anxiety, g = 0.38 for depressive symptoms, and in cardiovascular patients a blood pressure reduction of 3.08 mmHg without improvement in depression and anxiety.

Same measurement, different steeringTracking HRV daily
The devices compared head to headChest strap or wearable
Ring and band against the gold standardOura or Whoop
What the values are mostly used forZone 2 cardio
The marker training depends onZone 2 & VO2max
Breathing raises HRV acutelyBox breathing
Same sectionSauna & cold (hormesis)

Sources

Open in the database – with search, filters and comparison (German app)

Information only, not medical advice and not a usage recommendation. HRV does not replace an ECG or a medical evaluation. Last updated: 2026-10-04.