MVIIIHealth The MVIII app
← Back

Research · Devices and wearables

Heart rate variability wearables

Chest straps matched ECG for beat-to-beat timing at rest. Two meta-analyses found a small standardized error for portable devices at rest. In single studies, wrist and ring sensors differed from the reference by 6 to 31% on average depending on device, with less agreement while awake. Training guided by daily readings did not improve endurance performance more than fixed plans.

Checked 2026-11-13
Status
Sold in the US as general wellness products. FDA's January 2026 guidance names heart rate variability among the physiologic outputs a noninvasive sensing product may report as a general wellness product when the output is intended solely for wellness uses [1].
Measures
Beat-to-beat variation in the timing of heartbeats, most often reported as RMSSD in milliseconds, taken from optical pulse sensors on a finger or wrist or from electrical sensors in a chest strap.
Also called
HRV, HRV tracker, heart rate variability monitor, RMSSD, readiness score, recovery score, chest strap, Polar H10, Oura Ring HRV, WHOOP HRV, Apple Watch HRV, Garmin HRV
Used for
recovery, readiness, training load, stress, sleep quality, overtraining, illness detection, nervous system balance

Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on

Matched the reference method5

Reported risks0

No risks sourced yet.

How it compares with the reference method

Matched the reference method5

Limited

Heart rate variability at rest, Apple Watch on watchOS 4.2

One study found Apple Watch beat-to-beat intervals agreed with a chest strap at rest in 20 healthy volunteers, with gaps in the recording that lowered frequency measures.

The numbers and sources
Device
Apple Watch running watchOS 4.2, with beat-to-beat intervals taken from the Breathe app. The paper does not state the watch model.
Compared with
Polar H7 chest strap, not ECG
Effect
Reliability and agreement coefficients between the watch and the chest strap were above 0.9. The watch's interval series contained gaps, on average 5 per recording lasting 6.5 seconds each. Time-domain indices (SDNN, RMSSD, pNN50) were not significantly affected by the gaps, while low and high frequency power fell significantly [2].
Population
20 healthy volunteers during relaxation and mild mental stress [2].
Sources
[2]
Caveats
The reference was a chest strap, not ECG. Laboratory relaxation and mental stress tasks only. Published in 2018 on watchOS 4.2 [2].
Moderate

Heart rate variability at rest, Polar H10 chest strap

Shown to match ECG for beat-to-beat intervals and time-domain heart rate variability at rest in healthy adults across three studies, with error under 1% in the largest.

The numbers and sources
Device
Polar H10 chest strap
Compared with
Laboratory ECG [7][8]. Visual inspection of the raw ECG signal alongside a Holter monitor [9].
Effect
In 43 adults, agreement with ECG for mean RR interval, RMSSD and SDNN gave concordance of 0.99 or higher, mean absolute percentage error under 1% and limits of agreement no wider than -1.5 to 1.7 ms [7]. In 25 adults, resting values did not differ from ECG, and during a cycling ramp the bias in RR intervals was under 1 ms, while a nonlinear index (DFA a1) showed limits of agreement from -40.9 to 58.1% at high intensity [8]. In 10 adults across five activities, the strap correctly detected 99.6% of RR intervals overall and 99.4% during high-intensity activity [9].
Population
43 healthy adults aged 18 to 39 [7]. 25 recreationally active men and women [8]. 10 healthy individuals [9].
Sources
[7], [8], [9]
Caveats
Healthy young and active adults in short protocols. One study was supported by a grant from Polar Electro [9].
Limited

Heart rate variability at rest, Polar Vantage V2

One study found the wrist-worn Polar Vantage V2 slightly overestimated heart rate variability against a chest sensor in recreational athletes, with correlation of 0.93 overnight.

The numbers and sources
Device
Polar Vantage V2, wrist optical sensor
Compared with
A chest-worn heart rate sensor recording RR intervals, not ECG
Effect
Across 11,344 pooled intervals, the watch's pulse intervals differed from the chest sensor's RR intervals by a mean 0.17 ms (correlation 0.993). The watch overestimated LnRMSSD in the laboratory (mean bias 0.20, p < 0.001) and overnight (mean bias 0.17, p < 0.001). Overnight intraclass correlation was 0.931 for LnRMSSD and 0.998 for heart rate. The authors described accuracy as adequate and noted that overestimation mainly affected people with low values [10].
Population
Recreationally endurance-trained adults: 39 in a 5 minute supine laboratory recording and 29 in a 4 hour recording during sleep [10].
Sources
[10]
Caveats
The reference was a chest sensor, not ECG. Trained adults lying still or asleep [10].
Limited

Nighttime heart rate variability, Fitbit Versa 4

One pilot study of 8 people found nightly RMSSD from Fitbit Versa 4 did not differ significantly from an ambulatory ECG device.

The numbers and sources
Device
Fitbit Versa 4
Compared with
Firstbeat Bodyguard 3 ambulatory ECG
Effect
Median RMSSD was 21.66 ms on Fitbit and 20.56 ms on the ECG device (p = 0.39). In a mixed model, ECG values averaged about 92.3% of Fitbit values [15].
Population
8 participants wearing both devices for at least three nights [15].
Sources
[15]
Caveats
Eight participants, described by the authors as a pilot giving preliminary support [15].
Moderate

Nighttime heart rate variability, Oura Ring

Shown to track nightly average RMSSD closely against ECG in healthy adults across three studies, with larger errors over 5 minute windows and in older adults.

The numbers and sources
Device
Oura Ring, generation not stated [17][18]. Oura Ring Generation 3 [19].
Compared with
Chest ECG worn overnight
Effect
In 35 adults monitored at home, 5 minute RMSSD correlated 0.915 with ECG with a mean bias of -14.97 ms, and night averages showed lower error than 5 minute values. Low frequency power and the low to high frequency ratio had high error in both tests [17]. In 49 adults, nightly average heart rate variability agreed with ECG (r² 0.980, mean bias -1.2 ms), and 5 minute segments agreed less closely (r² 0.765) [18]. In 114 adults, agreement improved when at least 80% of beats in a segment were valid and values were averaged over 30 minutes or more, and more than half of participants aged 45 or older exceeded 10% median absolute percentage error for heart rate variability [19].
Population
35 healthy adults at home [17]. 49 adults aged 15 to 72 [18]. 92 adults under 45 and 22 aged 45 or older in two laboratory sleep studies [19].
Sources
[17], [18], [19]
Caveats
The lead author of one study was affiliated with Oura Health [18]. In two separate multi-device studies, Oura rings underestimated heart rate variability by 10.2 ms (Generation 2) and showed 6 to 7% error (Generations 3 and 4) [14][16].

Mixed results8

Limited

Heart rate variability at rest, Apple Watch Series 6

Mixed results: one study of 78 adults found Apple Watch Series 6 matched ECG for average beat intervals at rest but showed a 31% error for beat-to-beat variability.

The numbers and sources
Device
Apple Watch Series 6, recording through the Breathe app
Compared with
Biopac 3-lead ECG
Effect
At rest, mean absolute percentage error was 1.15% for R-R intervals and beats per minute, which the authors called near-perfect agreement. For the study's beat-to-beat variability measure, which it terms N-N intervals, error at rest was 31.31%, with moderate agreement. Agreement was highest at rest and lower while talking, watching a film and around walking [3].
Population
78 healthy adults aged 20 to 75, tested in a laboratory [3].
Sources
[3]
Caveats
Healthy volunteers in one laboratory session. The study received no external funding [3].
Limited

Heart rate variability at rest, Firstbeat Textile Strap, Oura Ring 2nd generation, Polar H10 and smartphone camera apps

Mixed results: one study of 5 adults found errors in resting RMSSD from 4% to 112% across seven device and app pairings.

The numbers and sources
Device
Polar H10 chest strap read through the Elite HRV and HRV4Training apps, Firstbeat Textile Strap, Oura Ring 2nd generation, and three iPhone 8 camera apps (Camera HRV, Elite HRV, HRV4Training)
Compared with
Shimmer3 multi-lead ECG
Effect
Mean absolute percentage error for RMSSD was 4.10% for Polar H10 with HRV4Training, 6.84% for Oura, 7.66% for Polar H10 with Elite HRV, 8.71% and 9.43% for the Elite HRV and HRV4Training camera readings, 11.27% for the Firstbeat strap and 112.36% for Camera HRV. Concordance correlation with ECG ranged from 0.98 (Polar H10 with HRV4Training) to 0.04 (Camera HRV), with Oura at 0.91 [5].
Population
5 healthy young adults contributing 148 resting trials [5].
Sources
[5]
Caveats
Five participants. The same chest strap gave different errors through two apps. The study received no external funding [5].
Limited

Heart rate variability at rest, Garmin vivoactive 4

Mixed results: one study in 263 people found Garmin pulse data matched ECG for longer-term variability measures and showed moderate concordance of 0.66 for RMSSD.

The numbers and sources
Device
Garmin vivoactive 4, using the watch's optical pulse signal analyzed by the researchers
Compared with
High-resolution (1000 Hz) ECG, 30 minutes under standardized conditions
Effect
Lin's concordance with ECG was 0.9998 for mean heart rate, 0.9617 for SDANN and 0.9613 for very low frequency power. Concordance was 0.6617 for rMSSD and 0.5919 for the short-term fractal exponent, which the authors called moderate. They advised caution with markers of short-term variability [6].
Population
104 survivors of heart attack, 129 patients after ischemic stroke and 30 controls [6].
Sources
[6]
Caveats
Mostly patients with cardiovascular disease, at rest. Variability was computed by the researchers from the pulse signal, so the figures do not describe the number the watch displays [6].
Strong

Heart rate variability at rest, portable and optical devices pooled

Mixed results: two meta-analyses found a small standardized error against ECG at rest, and their authors differed on whether that supports treating the readings as interchangeable.

The numbers and sources
Device
Portable heart rate variability devices pooled across 23 studies [11]. Optical pulse sensors, wearable and camera-based, pooled across 10 studies [12]. Wearable devices in 18 studies, 16 of them ECG-based straps [13].
Compared with
ECG
Effect
Across 23 studies and 301 effects, portable devices differed from ECG by an absolute standardized difference of 0.23 (95% CI 0.05 to 0.42, I² 78.6%). Error was larger for SDNN (0.44) and in tilt or recovery positions, and did not vary by device. The authors called the error small and acceptable [11]. For optical pulse sensors in healthy people, pooled absolute standardized error was 0.188 for RMSSD (95% CI 0.066 to 0.309) and 0.134 for SDNN (95% CI 0.014 to 0.255). The authors wrote that the estimates do not show interchangeability and do not extend to sleep, exercise, stress or free-living use [12]. A systematic review found correlation with ECG ranged from "very good to excellent" at rest and declined progressively as exercise intensity rose [13].
Population
23 studies published before July 2017 [11]. 43 studies of healthy or apparently healthy people, 10 with data for pooling [12]. 18 studies [13].
Sources
[11], [12], [13]
Caveats
Pooled studies were mostly at rest or under controlled conditions. Heterogeneity was high in one analysis [11]. The reviews pool device types and generations.
Limited

Nighttime heart rate variability, Apple Watch S6, Garmin Forerunner 245 Music, Oura Ring Generation 2, Polar Vantage V, Somfit and WHOOP 3.0

Mixed results: one study of 53 young adults found intraclass correlation with ECG from 0.24 to 0.99 across six devices, and every device underestimated heart rate variability.

The numbers and sources
Device
Apple Watch S6, Garmin Forerunner 245 Music, Oura Ring Generation 2, Polar Vantage V, Somfit, WHOOP 3.0
Compared with
ECG during or just before one night of laboratory sleep
Effect
Against ECG, WHOOP underestimated heart rate variability by 4.5 ms (intraclass correlation 0.99), Polar by 8.7 ms (0.65), Apple Watch by 9.6 ms (0.67), Oura by 10.2 ms (0.63), Somfit by 20.5 ms (0.69) and Garmin by 22.4 ms (0.24). For every device except WHOOP, readings were overestimated at low values and underestimated at high values [14].
Population
53 healthy adults (26 women), mean age 25.4, one night [14].
Sources
[14]
Caveats
The authors belong to a research group that receives funding and equipment from WHOOP Inc. and state WHOOP had no role in the study. Sampling periods differed by device, from 3 minutes near the start of sleep (Garmin) to 4 hours within sleep (Polar), and were not specified for Apple Watch and Oura. Apple Watch values were read through Sleep Watch, a third-party app [14].
Limited

Nighttime heart rate variability, Garmin Fenix 6, Oura Generation 3, Oura Generation 4, Polar Grit X Pro and WHOOP 4.0

Mixed results: one study of 13 adults over 536 nights found nightly heart rate variability error against ECG ranged from 6% to 16% across five devices.

The numbers and sources
Device
Garmin Fenix 6, Oura Generation 3, Oura Generation 4, Polar Grit X Pro, WHOOP 4.0
Compared with
ECG worn during sleep
Effect
Mean absolute percentage error for nightly heart rate variability was 5.96% for Oura Generation 4 (concordance 0.99), 7.15% for Oura Generation 3 (0.97), 8.17% for WHOOP 4.0 (0.94), 10.52% for Garmin Fenix 6 (0.87) and 16.32% for Polar Grit X Pro (0.82). The authors rated the Garmin and Polar concordance as "poor" and the WHOOP concordance as "moderate". For resting heart rate, error was 1.67% to 3.00% across the four devices analyzed [16].
Population
13 healthy adults (6 women), 536 nights in total [16].
Sources
[16]
Caveats
Thirteen participants. Funded by the US Air Force Research Laboratory, and the authors declared no competing interests. Firmware versions are not reported in the abstract [16].
Limited

Nighttime heart rate variability, WHOOP 2.0

Mixed results: one study of 6 adults found WHOOP 2.0 within 6% of ECG during deep sleep, an error as large as the changes it is used to detect.

The numbers and sources
Device
WHOOP 2.0
Compared with
ECG during slow-wave sleep
Effect
For the natural log of RMSSD, bias was 1.66% (SD 1.80) and limits of agreement plus or minus 5.93% with the company's own filter. These approached or exceeded the smallest worthwhile change (1.5 to 6.5%) and the coefficient of variation (3 to 13%) for the measure. Heart rate bias was 0.39% or less, well under its smallest worthwhile change [20].
Population
6 healthy young adults (3 women), mean age 22.9, over 15 sleep opportunities in a laboratory [20].
Sources
[20]
Caveats
Six participants. One author's position was later sponsored by WHOOP Inc., an arrangement the paper says began after data collection [20].
Limited

Heart rate variability over 24 hours, Samsung Gear Sport

Mixed results: one study of 28 adults found Samsung Gear Sport RMSSD correlated 0.78 with ECG during sleep and 0.25 while awake.

The numbers and sources
Device
Samsung Gear Sport smartwatch
Compared with
Shimmer3 chest ECG over 24 hours of daily life
Effect
In 5 minute windows during sleep, RMSSD correlated 0.778 with ECG with a mean bias of -18.24 ms and a mean absolute error of 22.44 ms. While awake, the correlation was 0.251 and the mean absolute error 32.11 ms. Heart rate correlated 0.941 with ECG during sleep and 0.675 while awake [21].
Population
28 adults (14 women) going about their daily routines for 24 hours [21].
Sources
[21]
Caveats
Variability was computed by the researchers from the watch's pulse signal. The authors concluded the watch gave acceptable time-domain values during sleep and not while awake [21].

Did not match the reference method1

Limited

Heart rate variability at rest, Apple Watch Series 9 and Ultra 2

One study found Apple Watch Series 9 and Ultra 2 underestimated heart rate variability by 8 ms on average against a chest strap, outside the study's 10 ms equivalence margin.

The numbers and sources
Device
Apple Watch Series 9 and Apple Watch Ultra 2
Compared with
Polar H10 chest strap analyzed with Kubios HRV software, not ECG
Effect
The watches underestimated heart rate variability by 8.31 ms on average (p = 0.025), with a mean absolute percentage error of 28.88% and a mean absolute error of 20.46 ms. Equivalence testing found the readings did not fall within the pre-specified margin of plus or minus 10 ms. Resting heart rate differed by -0.08 beats per minute, with a mean absolute percentage error of 5.91% [4].
Population
39 healthy adults providing 316 measurements over 14 days [4].
Sources
[4]
Caveats
The reference was a chest strap, not ECG. Funded by Science Foundation Ireland, and the authors declared no conflicts of interest [4].

Does using one change anything

Shown to help1

Moderate

Resting heart rate variability with training guided by daily readings

Shown to raise resting heart rate variability more than a fixed training plan in one meta-analysis, with no difference in resting heart rate.

The numbers and sources
Effect
Training guided by heart rate variability raised vagal-related heart rate variability indices more than predefined training (SMD 0.50, 95% CI 0.09 to 0.91). Resting heart rate did not differ (SMD 0.04, 95% CI -0.34 to 0.43) [23].
Population
Adults in endurance training trials [23].
Sources
[23]
Caveats
The outcome is the same measure used to steer the training. The authors note that index choice, recording position and baseline methods differed between trials [23].

Not shown to help1

Strong

Endurance performance with training guided by daily readings

Not shown to improve endurance performance or VO2 max more than a fixed training plan in two meta-analyses of small trials.

The numbers and sources
Effect
Across 8 studies, training adjusted to daily heart rate variability had no significant effect over predefined training on performance (g = 0.079, p = 0.597) or peak oxygen uptake (g = 0.171, p = 0.130), and a significant effect on submaximal physiological measures (g = 0.296, 95% CI 0.031 to 0.562). Most guided programs included fewer moderate or high intensity sessions [22]. A second meta-analysis found small, non-significant differences for maximal aerobic capacity (SMD 0.20, 95% CI -0.07 to 0.47) and endurance performance (SMD 0.20, 95% CI -0.09 to 0.48) [23].
Population
198 participants in 8 studies [22]. Trials of endurance training, the longest lasting 8 weeks [23].
Sources
[22], [23]
Caveats
Small trials using a range of recording devices and decision rules. One review reported fewer non-responders with guided training [22]. Its authors concluded that any advantage for group-level fitness or performance is by a small margin [23].

Sources

  1. US Food and Drug Administration. General Wellness: Policy for Low Risk Devices. Guidance for Industry and Food and Drug Administration Staff. January 6, 2026. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/general-wellness-policy-low-risk-devices
  2. Hernando D, Roca S, Sancho J, et al. Validation of the Apple Watch for Heart Rate Variability Measurements during Relax and Mental Stress in Healthy Subjects. Sensors. 2018. https://doi.org/10.3390/s18082619
  3. Bonneval L, Wing D, Sharp S, et al. Validity of Heart Rate Variability Measured with Apple Watch Series 6 Compared to Laboratory Measures. Sensors. 2025. https://doi.org/10.3390/s25082380
  4. O'Grady B, Lambe R, Baldwin M, et al. The Validity of Apple Watch Series 9 and Ultra 2 for Serial Measurements of Heart Rate Variability and Resting Heart Rate. Sensors. 2024. https://doi.org/10.3390/s24196220
  5. Stone JD, Ulman HK, Tran K, et al. Assessing the Accuracy of Popular Commercial Technologies That Measure Resting Heart Rate and Heart Rate Variability. Frontiers in Sports and Active Living. 2021. https://doi.org/10.3389/fspor.2021.585870
  6. Theurl F, Schreinlechner M, Sappler N, et al. Smartwatch-derived heart rate variability: a head-to-head comparison with the gold standard in cardiovascular disease. European Heart Journal Digital Health. 2023. https://doi.org/10.1093/ehjdh/ztad022
  7. Blalock ACE, Riemann BL, Flatt AA. Validity of the Polar H10 for heart rate variability and cardiac autonomic reflex tests. Autonomic Neuroscience. 2026. https://doi.org/10.1016/j.autneu.2026.103447
  8. Schaffarczyk M, Rogers B, Reer R, Gronwald T. Validity of the Polar H10 Sensor for Heart Rate Variability Analysis during Resting State and Incremental Exercise in Recreational Men and Women. Sensors. 2022. https://doi.org/10.3390/s22176536
  9. Gilgen-Ammann R, Schweizer T, Wyss T. RR interval signal quality of a heart rate monitor and an ECG Holter at rest and during exercise. European Journal of Applied Physiology. 2019. https://doi.org/10.1007/s00421-019-04142-5
  10. Nuuttila OP, Korhonen E, Laukkanen J, Kyröläinen H. Validity of the Wrist-Worn Polar Vantage V2 to Measure Heart Rate and Heart Rate Variability at Rest. Sensors. 2021. https://doi.org/10.3390/s22010137
  11. Dobbs WC, Fedewa MV, MacDonald HV, et al. The Accuracy of Acquiring Heart Rate Variability from Portable Devices: A Systematic Review and Meta-Analysis. Sports Medicine. 2019. https://doi.org/10.1007/s40279-019-01061-5
  12. Xu S, Liu H, Liu Z, et al. Accuracy of Photoplethysmography-Derived Pulse Rate Variability Compared with Electrocardiography-Derived Heart Rate Variability: A Systematic Review and Meta-Analysis. Sensors. 2026. https://doi.org/10.3390/s26165192
  13. Georgiou K, Larentzakis AV, Khamis NN, et al. Can Wearable Devices Accurately Measure Heart Rate Variability? A Systematic Review. Folia Medica. 2018. https://doi.org/10.2478/folmed-2018-0012
  14. Miller DJ, Sargent C, Roach GD. A Validation of Six Wearable Devices for Estimating Sleep, Heart Rate and Heart Rate Variability in Healthy Adults. Sensors. 2022. https://doi.org/10.3390/s22166317
  15. Nasr A, Behbehani K, Wang J, et al. Agreement between Fitbit Versa 4 and a reference standard on time-domain nocturnal heart rate variability: A pilot study. Digital Health. 2025. https://doi.org/10.1177/20552076251394960
  16. Dial MB, Hollander ME, Vatne EA, et al. Validation of nocturnal resting heart rate and heart rate variability in consumer wearables. Physiological Reports. 2025. https://doi.org/10.14814/phy2.70527
  17. Cao R, Azimi I, Sarhaddi F, et al. Accuracy Assessment of Oura Ring Nocturnal Heart Rate and Heart Rate Variability in Comparison With Electrocardiography in Time and Frequency Domains: Comprehensive Analysis. Journal of Medical Internet Research. 2022. https://doi.org/10.2196/27487
  18. Kinnunen H, Rantanen A, Kenttä T, Koskimäki H. Feasible assessment of recovery and cardiovascular health: accuracy of nocturnal HR and HRV assessed via ring PPG in comparison to medical grade ECG. Physiological Measurement. 2020. https://doi.org/10.1088/1361-6579/ab840a
  19. Liang T, Yilmaz G, Soon CS. Deriving Accurate Nocturnal Heart Rate, rMSSD and Frequency HRV from the Oura Ring. Sensors. 2024. https://doi.org/10.3390/s24237475
  20. Bellenger CR, Miller DJ, Halson SL, et al. Wrist-Based Photoplethysmography Assessment of Heart Rate and Heart Rate Variability: Validation of WHOOP. Sensors. 2021. https://doi.org/10.3390/s21103571
  21. Sarhaddi F, Kazemi K, Azimi I, et al. A comprehensive accuracy assessment of Samsung smartwatch heart rate and heart rate variability. PLOS ONE. 2022. https://doi.org/10.1371/journal.pone.0268361
  22. Düking P, Zinner C, Trabelsi K, et al. Monitoring and adapting endurance training on the basis of heart rate variability monitored by wearable technologies: A systematic review with meta-analysis. Journal of Science and Medicine in Sport. 2021. https://doi.org/10.1016/j.jsams.2021.04.012
  23. Manresa-Rocamora A, Sarabia JM, Javaloyes A, et al. Heart Rate Variability-Guided Training for Enhancing Cardiac-Vagal Modulation, Aerobic Fitness, and Endurance Performance: A Methodological Systematic Review with Meta-Analysis. International Journal of Environmental Research and Public Health. 2021. https://doi.org/10.3390/ijerph181910299

Everything in the library