HRV Biofeedback: What the Evidence Shows

What is HRV biofeedback?

HRV biofeedback is training in slow, paced breathing while watching your own heart rhythm on a screen in real time. It is a form of cardiorespiratory feedback training: the screen shows the heart speeding up on the in-breath and slowing on the out-breath, and the person learns to breathe so that these swings become large and smooth [S1].

The classic protocol has three parts:

• Find the resonance rate. A practitioner compares several slow breathing rates and picks the one that produces the greatest heart-rate oscillations [S2] — for most adults near about 5.5–6 breaths per minute (around 0.1 Hz).
• Breathe with feedback. In sessions, the person breathes at that rate while a sensor and a screen show the heart rhythm, so they can see what the breath is doing.
• Practise over weeks. Programmes typically combine a course of sessions with daily home practice; in pooled trials, the number of sessions and weeks did not clearly change the size of the effect [S3].

It has been studied for conditions as varied as asthma and depression, and for performance [S1]. Practical guidance lives in HRV biofeedback, find your resonance breathing rate and the resonance breathing guide; this page is about what the research shows.

How does it work?

Briefly: slow breathing near the resonance rate brings breathing, heart rate and blood pressure into step, and the blood-pressure reflex (the baroreflex) amplifies the heart-rate swings. The full mechanism is on how breathing changes HRV, and the breath-linked rhythm itself on respiratory sinus arrhythmia.

How this physiology might lead to less stress or better mood is a separate, less settled question. The most supported proposal involves training of the baroreceptor pathway and its links to the brain [S1] — a proposed mechanism, not an established one.

How is it measured?

Studies of HRV biofeedback measure two different things, and it helps to keep them apart:

• Outcomes — what the training is for: stress and anxiety questionnaires, depression scales, blood pressure, sleep ratings, sport or cognitive tests.
• Physiology — heart rate variability during practice and at rest, sometimes baroreflex measures.

A higher heart rate variability during slow breathing is expected from the breathing itself; it is not evidence that an outcome has improved. In one trial, higher breathing-related heart rate variability during practice did not systematically go with larger psychological benefits [S12], and a four-week programme did not change resting heart rate variability even though symptoms improved [S13].

The comparison condition matters as much as the measure. Effect sizes are larger against inactive controls — waiting lists, no training — than against active ones such as other relaxation methods, though significant for both [S3]. A true placebo is hard to build: participants know whether they are breathing slowly.

What affects it?

• The comparison. Inactive versus active control conditions change the size of the effect [S3].
• Population. Effects differ between healthy volunteers, people with anxiety or depression, patients with cardiovascular disease and athletes; the strongest psychological effects come from self-report in mostly small trials [S4].
• Protocol. Resonance assessment, session count, home-practice dose and equipment all differ between studies. In remote programmes, maximising resonance and having a screen on the device were among the features associated with better results [S6]; in cognitive studies, efficacy was not related to the duration or intensity of the intervention [S9].
• Questionnaire and timing. In the depression meta-analysis, the questionnaire used and the publication year moderated the effect [S5].

What does the evidence show?

Overall: established, with a modest size. A systematic review and meta-analysis of 58 randomized controlled studies across many outcomes found a small to moderate effect favouring HRV biofeedback, similar to that of other effective treatments [S3]. The authors describe it as useful as a complementary treatment and call for more research for particular applications [S3].

Stress and anxiety: context-dependent. A meta-analysis of 24 studies with 484 participants found that the training was associated with a large reduction in self-reported stress and anxiety, while noting that more well-controlled studies are needed [S4]. A newer meta-analysis of remotely delivered programmes did not find a significant effect on stress [S6]. The benefit is clearest for self-reported symptoms in small trials.

Depressive symptoms: context-dependent. A meta-analysis of 14 randomized controlled trials with 794 participants in adults found a medium mean effect, with moderate heterogeneity and a prediction interval that crosses zero [S5]. Remote programmes showed a medium effect on depression as well [S6]. In patients with cardiovascular disease, by contrast, depression and anxiety did not improve significantly [S7].

Blood pressure: emerging. In patients with cardiovascular disease, a meta-analysis of 13 randomized controlled trials with 965 participants found modest decreases in systolic and diastolic blood pressure, with most trials at some or high risk of bias [S7]. Practical angles are in slow breathing and high blood pressure.

Sleep: emerging, small. Sleep is among the outcomes with the smallest effects in the broad meta-analysis [S3], and positive sleep findings in chronic-disease studies come from a narrative review without pooled estimates [S10]. No dedicated sleep meta-analysis was found.

Sport and cognitive performance: emerging. A systematic review in sport found only a few studies, all with small samples, most reporting benefits [S8]. In cognitive studies, about half reported a significant benefit on at least one executive function, most often attention [S9].

Other conditions: emerging. A systematic review in chronic disease reported that the training was feasible without adverse effects and found positive results across hypertension, asthma, depression and anxiety, sleep disturbances, cognitive performance and pain, and asked for confirmation [S10].

Does the feedback itself add anything? Debated. This is the central open question. Slow breathing is associated with higher vagally mediated HRV; whether a longer exhale adds anything beyond slowing the breath is still debated. Without any feedback screen, a large meta-analysis found these increases during practice, right after a single session and after multi-session programmes [S11]. Trials that isolate the screen are scarce:

• Against a sham version of biofeedback, real biofeedback improved positive affect and reduced depression scores in healthy adults, but did not change autonomic measures [S12]. The abstract does not say exactly what the sham kept or removed.
• In one randomized trial, biofeedback at an individually selected resonance rate did not outperform biofeedback at a fixed slow rate; the authors say their trial does not permit firm conclusions [S13]. This is the result of a single study, not a settled answer.
• Remote programmes with a screen on the device tended to do better, but that is an association across studies, not a test [S6].

The most defensible reading today: the slow breathing likely does much of the physiological work; the feedback may help people learn the technique and keep practising, but how much it adds on its own is not settled. The effects of slow breathing without feedback are covered in slow breathing: what the evidence shows.

What remains uncertain.

• Whether the psychological effects hold in large, well-controlled trials with credible active or sham comparisons [S3] [S4] [S12].
• How much the on-screen feedback adds beyond slow breathing [S11] [S12] [S13].
• Whether individual resonance assessment matters for outcomes; so far one randomized trial found no advantage [S13].
• How long benefits last after practice stops.

What it does not tell you

• It is not a treatment on its own. The research describes HRV biofeedback as a complementary method [S3]. It does not replace psychotherapy, medication or medical care for anxiety, depression, high blood pressure or any other clinical condition, and nothing here means "stop your treatment".
• A higher number during practice is not proof of benefit. Heart rate variability rises with slow breathing by design [S11]; psychological gains did not track it closely in one trial [S12], and resting values did not change in another [S13].
• The effect sizes are not final. Most trials are small, protocols differ, blinding is hard and many studies carry a risk of bias [S7] [S8] [S9]; estimates may shrink as better trials arrive.
• It is not a readout of vagal tone. Vagal tone cannot be measured directly; HRV measures such as RMSSD reflect vagally mediated changes in heart rate.
• Laboratory results may not carry over to wearables. Most trials used laboratory sensors and software. Watches, rings and phone cameras estimate heart rhythm from the pulse, which is related to but not identical with the electrical signal [S14] — see can you trust HRV from a smartwatch. Whether biofeedback delivered through these devices produces the same outcomes has not been tested directly; the meta-analysis of remote programmes is the closest evidence [S6].
• Symptoms need a clinician. Chest pain, fainting, severe breathlessness or palpitations with dizziness need medical attention, not a breathing session. For how breathing practice compares with meditation, see meditation vs breathwork.

In ONDA

ONDA is a guided-breathing app with real-time heart-rhythm feedback. During practice, the iPhone camera shows live pulse and a breathing-rate estimate, and an Apple Watch adds a live coherence score — ONDA's own measure of how smooth and rhythmic the heart-rhythm oscillation is, not a clinical heart-rate-variability measurement. Guidance is spoken and visual rather than a fixed numeric breathing pacer, so ONDA is not the laboratory protocol tested in the studies above. ONDA has no study of its own effectiveness, and the findings on this page do not show what ONDA does or does not achieve. ONDA does not diagnose or treat any condition. See Apple Watch HRV biofeedback and what ONDA measures.