Altitude and HRV: What Changes and What It Cannot Predict

What happens to HRV at altitude?

At high altitude each breath brings in less oxygen. Heart rate variability (HRV) is the beat-to-beat variation in the interval between heartbeats, and in the first days at altitude it usually falls. The most reliable evidence is a meta-analysis of 15 studies with 698 participants: healthy adults who went up to 2,500 m or higher and were measured within their first seven days there [S1]. Compared with sea level, SDNN, RMSSD, high-frequency power and related HRV measures were all lower [S1].

The authors interpret the pattern as vagal withdrawal with relative sympathetic predominance [S1]. The framing matters. Vagal tone cannot be measured directly; HRV measures such as RMSSD reflect vagally mediated changes in heart rate, so this is a reading of measured HRV, not a direct measurement of nerve activity. Part of that reading rests on a rise in the ratio of low- to high-frequency power (LF/HF). This ratio is a disputed measure of sympathetic–parasympathetic balance. In their own discussion, the authors add that this ratio mainly reflects how the remaining, smaller spectral power is redistributed, rather than a direct measure of sympathetic tone [S1]. Why this ratio is debated is explained on the autonomic nervous system.

Part of the drop may be mechanical. Low oxygen speeds up breathing, and the authors relate part of the fall in high-frequency power to that faster breathing [S1]. So the change reflects how you breathe as well as how the heart is regulated.

Being fit did not prevent the drop: HRV fell in trained people as well as in other healthy adults [S1]. The pooled studies have clear limits. The highest altitude among them was 4,000 m, recordings were short clinical ECGs rather than wearables, most participants were young men, and the authors note limited data for women, older adults and less fit people [S1].

Is the drop stronger higher up?

Only for some measures. The meta-analysis split the studies at about 3,500 m [S1]. In the higher group, SDNN, a measure of overall variability, fell further [S1]. RMSSD and high-frequency power, the measures most closely tied to breathing-linked, vagally mediated changes, did not differ significantly between the lower- and higher-altitude studies [S1]. The authors interpret this part of the response as early-onset and early-plateauing [S1]. Because this compares different studies rather than the same people at different altitudes, it does not show that RMSSD stops falling higher up; it only shows no significant difference between the two groups of studies. RMSSD itself is explained on RMSSD.

Can HRV predict altitude sickness?

Acute mountain sickness (AMS) is the common form of altitude sickness. Whether HRV can predict it is contested: the reviewers themselves note that results from individual studies have been inconsistent [S2]. Both sides are set out below; the evidence does not support a verdict either way.

A meta-analysis that found modest associations. The only meta-analysis on the question included 7 studies with 329 participants, at altitudes between 2,400 and 5,085 m, on real mountains and in altitude chambers [S2]. People who later developed AMS had a higher pNN50, the share of successive beat-to-beat intervals that differ by more than 50 ms, before the ascent (standardized mean difference 0.40, 95% CI 0.11 to 0.69) [S2]. That is a higher vagally linked value, not a lower one. Those who had AMS showed a lower SDNN after the ascent (standardized mean difference −0.41, 95% CI −0.69 to −0.13) [S2]. Each association rests on only part of the review: the first on two studies with 192 participants, all men, the second on three studies with 204 participants. Low- and high-frequency power showed no significant difference [S2]. The authors conclude that further research is needed to establish definitive clinical guidelines [S2].

A field study that found a signal at one altitude. In a study of 36 healthy lowland climbers (34 men and 2 women), measured every morning on high-mountain expeditions as they climbed from 2,400 m, RMSSD and high-frequency power, from a short chest-strap recording taken lying down, were 17–51% lower at 2,400 m in the climbers who went on to develop acute mountain sickness at 3,000–4,300 m than in the others [S3]. The climbers were grouped afterwards into three groups of 12 climbers: those who developed acute mountain sickness at 3,000–4,300 m, those who developed it only at 5,000 m or higher, and those who never did [S3]. In this study, an RMSSD of 30 ms or less and, separately, an oxygen saturation of 91% or less measured straight after exercise, both at 2,400 m, each picked out 92% of the climbers who developed acute mountain sickness at 3,000–4,300 m [S3]. No combined rule was tested. The same cut-offs also flagged many climbers who stayed well, with a specificity of 58% for the RMSSD cut-off and 40% for the oxygen-saturation cut-off [S3]. The early signal also missed later illness: average RMSSD at the first measurement altitude was 21 ms in the climbers who got acute mountain sickness at 3,000–4,300 m, 43 ms in those who never got it and 48 ms in those who got it only at 5,000 m or higher [S3]. The cut-offs were set in the same small sample and never tested in new people, the ascents were faster than generally recommended, and the authors call the correlations between HRV and AMS rather weak [S3]. This is a single study, and it is also one of the studies in the meta-analysis above [S2], so the two are not independent confirmations.

A study that found no prediction. In a study of 16 healthy British servicemen whose heartbeats were recorded with a chest patch on eight consecutive nights, starting at 800 m and sleeping at up to 3,600 m, AMS occurred in 7 of the 16 participants (43.8%) and was very mild in 85.7% of cases [S4]. The authors report: "HRV failed to predict AMS." [S4] With few and mostly very mild cases, this single study had little room to detect a prediction, and the authors cannot be certain the finding holds at higher altitudes or with more severe illness [S4]. The company that made the patch supplied and paid for the patches, and two of the authors worked for it.

Where this leaves the question. The studies point in different directions, and none tested whether a reading can warn an individual. None of these studies shows that HRV, from a watch or any other device, can warn an individual of altitude sickness.

Does HRV return to normal once you acclimatise?

A common belief is that you acclimatise to altitude in a couple of days. None of the studies on this page tests acclimatisation as a whole, with breathing, oxygen levels and symptoms, so for that belief the answer here is: not shown. What one small study did test is narrower: whether the resting heart-rhythm pattern returns to its sea-level form during a long stay at very high altitude.

In a study of seven people aged 27–35, measured at sea level and during a 35-day stay at 5,050 m, and six Sherpas aged 22–30, measured at altitude only, the share of high-frequency, breathing-linked power measured lying down fell from 25% to 10.9% of total spectral power [S5]. Sitting, no change was seen at altitude; at sea level, sitting up had already lowered that share [S5]. The authors conclude that ten days above 2,850 m followed by a month at 5,050 m did not change this pattern [S5]. The Sherpas, who live at altitude, showed comparable results [S5].

So, in this one study at very high altitude, the breathing-linked share of the heart rhythm had not returned to its sea-level pattern after the long stay. It is a single small study from 1996 at one altitude, using spectral measures only. It reports no RMSSD, says nothing about moderate altitudes or about symptoms, and its full text could not be checked for conflicts of interest.

A clinical sample. In a chamber study of 10 men with coronary artery disease who slept either at a simulated 1,900 m or at 250 m before being taken to a simulated 3,000 m the next morning, the night at the intermediate altitude was followed by measurable ventilatory acclimatisation on the next ascent, while HRV and baroreflex sensitivity showed no differences [S11]. So breathing can start to adjust while the heart-rhythm measures do not. This is a single small study in men with heart disease; it does not apply to healthy people and is not advice on ascent plans.

What about people who live or work at altitude?

Everything above concerns lowlanders who travel up. People who live at altitude are a different case. In a cross-sectional study of 405 healthy native residents of Sikkim, in the Himalayas, living below 1,500 m, between 1,500 and 2,500 m, or above 2,500 m, those living higher up had higher SDNN and RMSSD [S7]. The authors read this as a possible adaptive response to long-term low oxygen [S7]. In Bogotá, a study compared adults raised at 2,500 m or higher with adults raised below 1,500 m, all living in Bogotá, at 2,600 m, for more than six months: adults raised at low altitudes showed lower HRV than those raised at high altitudes [S8]. The authors relate this to altitude in early life rather than to a direct causal effect of low oxygen in childhood [S8].

Both are single cross-sectional studies. They compare different people at one point in time, cannot separate genes, upbringing and surroundings, and do not show how a visitor's HRV changes over months. The picture is not uniform either: in the small 1996 study, the Sherpas showed results comparable to the lowlanders [S5]. None of this applies to a trip to the mountains.

An occupational group. In a study of 19 male mine workers aged 35–55 with more than 10 years of alternating weeks at sea level and at about 3,800 m, recorded for four hours each night, RMSSD during those hours fell at altitude while SDNN rose [S9], so different HRV measures can move in opposite directions. These workers have spent years alternating weeks at altitude and at sea level, which is not the same as a first trip; it is a single study in men.

Can a wearable measure HRV at altitude?

It can record HRV, but the fast, breathing-linked details are less reliable than an ECG. A method study compared a finger pulse sensor, which reads the pulse optically (photoplethysmography, PPG), with an ECG during sleep in 21 volunteers sleeping at 4,554 m and five mountain guides sleeping at sea level, 6,000 m and 6,800 m [S6]; four of the five guides were recorded at the highest camp [S6]. The pulse sensor gave usable HRV even there [S6].

The average interval between beats from the pulse sensor was within a few milliseconds of the ECG (815 vs 813 ms) [S6]. The faster, breathing-linked components were not: even the closest pulse-timing method read high-frequency power up to 32% higher than the ECG, and the least accurate one read it at more than twice the ECG value [S6]. More pulse data had to be discarded as artefact, 0.1–5.5% of the finger-sensor signal, against at most 0.2% of the ECG [S6]. In the guides, the pulse sensor showed a significant rise in the LF/HF ratio that the ECG did not show [S6]. A pulse sensor can therefore exaggerate or hide exactly the kind of shift this page describes. The authors explain the gap by changes in how fast the pulse wave travels [S6].

The limits are important. The sensor was a finger clip connected to a research ECG vest, not a watch or a ring, and the authors note that sensors at other sites, such as the wrist, may give different results [S6]. The researchers chose stable stretches of sleep by eye and cleaned artefacts by hand [S6], which consumer devices do not do, and the results at the highest camps rest on very few people. This single study is used here only for how well the sensors agree, not for the direction of the altitude effect. How wearable HRV is measured is covered on heart rate variability.

What can you expect from your own numbers at altitude?

In the first days at altitude, lower HRV is common: in pooled short ECG recordings, HRV was lower than at sea level [S1]. The path is not always straight. In the field study of climbers, RMSSD in those who stayed well first rose in the first days of the ascent; above about 3,500 m all HRV measures fell while heart rate rose [S3]. How resting pulse is measured and what moves it is covered on resting heart rate.

A drop is common, but it does not always show. In the servicemen's study, HRV was analysed for one hour each night, between 2 and 3 a.m.. None of the time-domain measures, such as RMSSD, changed significantly during sleep [S4], but the altitude was moderate, and only eight of them slept at the highest altitude, about 3,660 m, for two nights; the other eight slept at about 2,540 m. In the same study, how hard the previous day had felt was linked with the next night's HRV [S4], so the day's effort moves the night's numbers too (exercise and HRV).

Recovery after hard exercise is a separate question from resting HRV. In a trial of 12 recreationally trained people who did two repeated-sprint formats, in normal air and in simulated low oxygen, the sprint format, not the low oxygen, changed HRV in the first minutes of recovery [S10]. That concerns recovery after sprints, not resting HRV in the first days at altitude, which the meta-analysis above describes [S1].

Comparing these nights with your usual baseline from home says little. A gap is expected, and it does not tell you whether you are adjusting well or becoming ill. More useful is the trend over several days at the same altitude, read together with how you feel. A single low HRV reading does not by itself mean you are stressed or unwell. Short sleep, alcohol, hard effort and infection also move these numbers (why HRV changes from day to day; sleep and HRV; illness and HRV; interpreting HRV).

What does the evidence show?

By evidence class.

• Context-dependent. In healthy adults in their first days at altitude, HRV is lower than at sea level, in trained and untrained people alike [S1]. Higher up, SDNN falls further, while RMSSD and high-frequency power do not differ significantly between lower- and higher-altitude studies [S1]. Part of the high-frequency fall may reflect faster breathing [S1].
• Emerging. A finger pulse sensor records HRV at very high altitude but diverges from the ECG on the fast components [S6]. In one small study at very high altitude, the breathing-linked share of the heart rhythm stayed low after a long stay [S5]. Night-time time-domain HRV did not change significantly at moderate altitude in one study [S4]. People who live at altitude differ from visitors: residents living higher had higher HRV [S7], and adults raised at low altitude had lower HRV in a high-altitude city than those raised there [S8]. In an occupational group, night-time RMSSD fell at altitude while SDNN rose [S9]. In a clinical sample, a night at moderate altitude changed breathing but not HRV on the next ascent [S11]. After sprints, the format rather than low oxygen changed early recovery HRV [S10].
• Debated. Whether HRV predicts acute mountain sickness: modest associations in a meta-analysis [S2] and a signal at one altitude in a field study [S3], against no prediction in another study [S4]. The reading of the shift as sympathetic predominance, which rests partly on LF/HF [S1].
• Unknown or not shown. That HRV from a watch can warn an individual of altitude sickness; that you acclimatise in a couple of days; how wrist sensors perform at altitude.

Safety

This is the most important part of the page. HRV and watches do not replace checking for symptoms of altitude sickness.

• Symptoms decide, not the watch. Headache, nausea, breathlessness at rest, confusion, unusual drowsiness or unsteadiness at altitude are reasons to stop going higher, to descend and to seek medical help.
• Confusion, severe breathlessness, chest pain or fainting need emergency help.
• A normal-looking reading does not rule out altitude sickness. In the studies above, HRV did not reliably identify who became ill: one missed those who got sick higher up [S3] and another found no prediction [S4].
• Do not delay descent to wait for a better reading.
• This page gives no advice on ascent plans, acclimatisation schedules or medicines.

What it does not tell you

• HRV is not a test for altitude sickness. The studies disagree, and none shows that a reading can warn an individual [S2] [S3] [S4].
• A drop at altitude is expected, not damage. It is the usual response in the first days [S1].
• Your baseline from home is the wrong yardstick in the first days. The gap is expected and does not show how well you are adjusting.
• Residents are not visitors. Findings in people who live at altitude [S7] [S8] do not show what happens on a trip.
• Most of the evidence comes from young, fit men. Data for women, older adults and less fit people are limited [S1].
• The LF/HF ratio is debated. Even the authors who report its rise add that it is not a direct measure of sympathetic tone [S1].
• No medical conclusions. None of these studies diagnoses a condition or gives an ascent plan.

In ONDA

ONDA is built around practice rather than tracking: it offers guided breathing practices with spoken and visual guidance, and ONDA shows your pulse live during a practice, from the iPhone camera or an Apple Watch. With an Apple Watch, or a device that syncs heart data to Apple Health, it compares your HRV, resting heart rate and breathing rate with your own baseline over 14 days. Apple Health records HRV as SDNN, so ONDA's HRV trend is an SDNN trend. In the first days after you travel to altitude, that baseline still reflects your nights at home, so a signal then says that your nights differ from home, which is expected at altitude, not that you are getting altitude sickness. Because the 14 days baseline is rolling, it gradually takes in the altitude nights, so what it compares against changes during the stay. ONDA does not assess or warn of altitude sickness, does not diagnose any condition and does not replace a doctor; the studies on this page did not test ONDA. See what ONDA measures.