How to Interpret HRV in Clinical Practice: Measurements, Limitations, and Applications

Author: Lisette Portes, MD, Board-Certified Family Medicine Physician

Published / Last Reviewed: November 3, 2026 / Medically reviewed by Lisette Portes, MD

HRV interpretation in clinical practice depends on method. The 1996 Task Force standards recommend 5-minute stationary recordings for spectral analysis and nominal 24-hour recordings for time-domain analysis, and the two should not be compared. RMSSD and HF power best reflect vagal modulation, LF/HF is not a reliable sympathovagal index, and HRV is a risk marker rather than a diagnosis. See our HRV overview and normal HRV ranges.

What Are the Core Principles of HRV Interpretation?

HRV interpretation rests on one principle: a value is meaningful only when the recording method, duration, and conditions are known and matched. Heart rate variability is the variation in normal-to-normal (NN) intervals between sinus beats, so it reflects autonomic modulation of the sinus node, not nerve traffic itself. The 1996 Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, published in Circulation, recommended two standard recordings. These are short-term recordings of 5 minutes under physiologically stable conditions for frequency-domain analysis, and nominal 24-hour recordings for time-domain analysis. The Task Force stated that it is inappropriate to compare time-domain measures from recordings of different durations. Shaffer and Ginsberg, in a 2017 Frontiers in Public Health review, similarly cautioned that 24-hour, short-term, and ultra-short-term norms are not interchangeable. For a patient-friendly primer, see what HRV is.

How Should Clinicians Read Time-Domain HRV Metrics Such as SDNN, RMSSD, and pNN50?

Time-domain metrics describe beat-to-beat variation in milliseconds, and RMSSD is the primary time-domain index of vagally mediated change. SDNN, the standard deviation of NN intervals, reflects total variability and depends heavily on recording length. RMSSD is the root mean square of successive differences, and pNN50 is the percentage of adjacent intervals differing by more than 50 ms. The Task Force listed approximate 24-hour normal values from small studies not adjusted for age or sex: SDNN 141 plus or minus 39 ms, RMSSD 27 plus or minus 12 ms, and HRV triangular index 37 plus or minus 15. For short-term resting recordings, a 2010 review by Nunan and colleagues pooled 44 studies with 21,438 healthy adults and reported mean SDNN of 50 ms (range 32 to 93) and mean RMSSD of 42 ms (range 19 to 75). Laborde and colleagues (2017) recommend RMSSD over pNN50 as a vagal index.

What Do Frequency-Domain and Nonlinear HRV Measures Really Show?

Frequency-domain HRV divides variability into bands, and only high-frequency (HF) power has a reasonably specific physiological meaning, reflecting respiratory-linked vagal modulation. Shaffer and Ginsberg list the bands as VLF 0.0033 to 0.04 Hz, LF 0.04 to 0.15 Hz, and HF 0.15 to 0.40 Hz. The Task Force reported 5-minute supine values of LF 1170 plus or minus 416 ms squared and HF 975 plus or minus 203 ms squared, with LF/HF of 1.5 to 2.0. The interpretation of LF has changed since then. Shaffer and Ginsberg state that at rest the LF band reflects baroreflex activity rather than cardiac sympathetic innervation, and they cite Billman's argument that about half of LF variability is parasympathetic. Laborde and colleagues therefore discourage using LF/HF as a sympathovagal balance index. Among nonlinear measures, the Poincare SD1 is mathematically equivalent to RMSSD, while SD2 reflects longer-term variability and correlates with LF power.

How Do Recording Length, Conditions, and Artifacts Change HRV Results?

Recording length, posture, timing, breathing, and artifact handling can change HRV more than many clinical variables, so standardization is essential. Laborde and colleagues (2017) recommend 5-minute recordings when possible, at least 1 minute for HF analysis, and note that 10-second segments can yield valid RMSSD, with multiple segments preferred. Respiratory rate should stay within 9 to 24 breaths per minute, matching the 0.15 to 0.40 Hz HF band, for HF to be read as vagal tone. The Task Force required manual editing of RR data to a very high standard and warned that automatic filters, such as excluding intervals that differ by more than 20% from the previous one, should not replace manual review. Laborde and colleagues describe an example in which an automatic correction filter produced an 11% error in the vagal estimate.

Record at a consistent time of day, after at least 5 minutes in the test posture.
No meal, coffee, tea, or energy drinks in the 2 hours before recording.
No alcohol for 24 hours and no intense training the day before.
Inspect the ECG visually and document every edited beat.
Average at least two resting measurements when using HRV as a trait.

How Should Clinicians Handle Confounders and Consumer Wearable HRV Data?

Age, sex, heart rate, medications, rhythm, and device type all confound HRV and must be considered before attributing a change to autonomic status. HRV declines with age, and Shaffer and Ginsberg note that in the Umetani dataset, values after age 65 fell below cutoffs linked to increased mortality, while sex differences seen before age 30 disappeared after 50. Sacha (2013) showed that HRV depends on mean heart rate partly for mathematical reasons, because RR interval and heart rate are inversely and nonlinearly related, and proposed normalizing HRV to the mean RR interval. Medications matter. In a 1994 study of 20 post-MI patients, Sandrone and colleagues found that atenolol or metoprolol increased 24-hour RR variance and normalized HF power. Kemp and colleagues (2010) found that tricyclic antidepressants reduce HRV. Atrial fibrillation, frequent ectopy, and pacing make standard HRV uninterpretable.

A 2018 systematic review by Georgiou and colleagues found PPG wearables usable only at rest or during mild exercise.
Accuracy falls as exercise intensity rises because of motion artifact.
A 2022 validation by Cao and colleagues found Oura nocturnal RMSSD accurate, but LF and LF:HF had high error.
Pulse rate variability from PPG is not identical to ECG-derived HRV.

What Should Clinicians Know About the Validated Clinical Applications of HRV?

The strongest clinical evidence for HRV is in post-myocardial infarction risk stratification and detection of diabetic cardiovascular autonomic neuropathy. In 1987, Kleiger and colleagues reported in the American Journal of Cardiology that a 24-hour SDNN under 50 ms after acute MI predicted markedly higher mortality, and Shaffer and Ginsberg summarize the mortality risk as 5.3 times higher than with SDNN over 100 ms. The Task Force defined highly depressed HRV as SDNN under 50 ms and HRV triangular index under 15, and moderately depressed HRV as SDNN under 100 ms and triangular index under 20, while noting that HRV alone has modest predictive value. For diabetes, the Task Force called depressed HRV an early warning sign of neuropathy. Cardiovascular reflex tests, including heart rate response to deep breathing, standing, and Valsalva, plus blood pressure on standing, remain the diagnostic standard under the 2011 Toronto consensus.

Primary prevention: a 2013 Hillebrand meta-analysis linked low HRV with a 32% to 45% higher risk of a first cardiovascular event.
Biofeedback: a 2017 meta-analysis by Goessl and colleagues of 24 studies found a Hedges g of 0.83 versus control for self-reported stress and anxiety.
Depression: Kemp and colleagues (2010) found lower HRV that tracked with severity.
Toronto staging: one abnormal cardiovagal test is possible CAN, two or more is definite.

How Can an Integrative, Physiology-Focused Doctor Help Patients Use HRV Data?

A practical framework is to treat HRV as a within-person trend, interpreted against symptoms, sleep, training, illness, and medications, rather than as a single score compared with population norms. Ask whether the recording method was consistent, whether rhythm was sinus, and whether a change persists across several days before acting on it. Dr. Lisette Portes is a board-certified family medicine physician in Miami who uses an autonomic, physiology-focused perspective when evaluating patients. Patients can discuss wearable HRV data, dizziness, fatigue, sleep, stress, and chronic conditions with her, in English or Spanish, in person or by telehealth for those located in Florida. Learn about chronic disease management or contact the office at 305-328-5079. Call 911 for chest pain, fainting with injury, new irregular heartbeat with symptoms, or stroke signs. This article is general education, not medical advice.

Frequently Asked Questions About HRV Interpretation

What is a normal RMSSD for adults?

There is no single normal RMSSD. A 2010 review by Nunan and colleagues, covering 44 studies and 21,438 healthy adults, reported a mean short-term resting RMSSD of 42 ms, with a range of 19 to 75 ms. Values fall with age and vary by method, posture, and device, so personal trends are more useful than one reading. See our HRV by age guide.

Is the LF/HF ratio a measure of sympathovagal balance?

Not reliably. The LF/HF ratio was originally proposed as a sympathovagal balance index, but current evidence shows that resting LF power largely reflects baroreflex and parasympathetic activity rather than cardiac sympathetic tone. Shaffer and Ginsberg (2017) summarize this critique, and Laborde and colleagues (2017) discourage using LF/HF for that purpose. RMSSD and HF power are better supported markers of vagal modulation.

How long should an HRV recording be for clinical use?

The 1996 Task Force standards recommend 5-minute recordings under stable conditions for frequency-domain analysis and nominal 24-hour recordings for time-domain analysis such as SDNN. Ultra-short recordings can estimate RMSSD, and Laborde and colleagues note that 10-second segments may be valid when several are averaged. Results from different recording lengths should never be compared directly.

Do beta-blockers affect heart rate variability?

Yes. In a 1994 study of 20 patients after a first myocardial infarction, Sandrone and colleagues found that atenolol or metoprolol slowed heart rate, increased 24-hour RR variance, and raised normalized high-frequency power. Because the drugs change HRV directly, clinicians should document beta-blocker use and timing, and compare a patient's readings only under similar medication conditions.

Can HRV be measured accurately in atrial fibrillation?

Standard HRV analysis is not valid in atrial fibrillation. HRV is defined by normal-to-normal intervals between sinus beats, and the Task Force standards assume sinus rhythm with careful editing of ectopic beats. In atrial fibrillation, irregular ventricular responses reflect the arrhythmia rather than autonomic modulation of the sinus node. Frequent ectopy and paced rhythms create similar problems for interpretation.

Are wearable HRV readings accurate enough for clinical decisions?

Wearables can track resting or overnight RMSSD trends reasonably well, but they are not diagnostic tools. A 2018 systematic review by Georgiou and colleagues found that PPG devices work only at rest or during mild exercise, and a 2022 Oura validation found accurate nocturnal RMSSD but high error for LF and LF:HF. Use wearable data as context alongside clinical evaluation.

HRV Interpretation in Clinical Practice: Metrics, Pitfalls, and Evidence-Based Uses