Stress monitoring devices and apps compared

Short answer

Garmin, Oura, WHOOP, Samsung and Google all sell a stress number. Apple does not. Every one of those numbers comes from heart rate and heart-rate variability, none has regulatory clearance as a medical device, and none has a published validation of the stress algorithm itself. The one pre-registered head-to-head test found a stress score tracked heart rate closely and perceived stress not at all.

If you have looked at a stress score and thought it did not match your day, you were not imagining it. That does not make the numbers useless. It makes them a measure of physiological arousal, which is not the same thing as stress, and worth reading accordingly.

Note: this is a wellness comparison, not medical advice. None of these products is a diagnostic tool. Statements about other companies come from their own documentation, support pages, developer docs and regulatory filings read on 22 September 2026, and from named studies.

Do you need a device to track stress?

No. Every stress monitoring device in this comparison computes its score from beat-to-beat intervals. A phone camera captures those from a fingertip using photoplethysmography — the same optical principle as a wrist sensor or a ring.

Welltory's camera measurement was compared against simultaneous ECG in 26 professional cyclists and published in Computer Methods and Programs in Biomedicine in 2022, with correlations of r = 0.77–0.94.

A wearable gives you continuous background sampling and the overnight window. A phone gives you a spot measurement whenever you choose. For stress specifically, the spot measurement has an underrated advantage: you control the conditions. Continuous scoring has to guess whether your heart rate rose because you were anxious or because you climbed stairs, and as you will see below, that guess is where most of the error lives.

What each one actually reports

Garmin — Stress Level, 0–100. Garmin's manual says the device "analyzes your heart rate variability while you are inactive to determine your overall stress." Bands: 0–25 resting, 26–50 low, 51–75 medium, 76–100 high. Grey bars appear when you were "too active to determine your stress level." Body Battery draws on HRV, the stress score, sleep quality and activity.

Which HRV statistic Garmin uses, at what sampling rate, is not in Garmin's consumer documentation. The underlying method comes from Firstbeat, a Garmin subsidiary, whose white paper is the only primary technical description: RR intervals resampled at 5 Hz, band-passed at 0.03–1.2 Hz, from which RMSSD, high-frequency power, low-frequency power and respiratory sinus arrhythmia amplitude are computed, then classified by a decision tree. Firstbeat defines stress as elevated heart rate with HRV below the individual's resting level and a low respiration rate relative to heart rate.

Oura — Daytime Stress and Resilience. Inputs are heart rate, HRV, motion and body temperature. Oura is the only company here that publishes its sampling interval: the graph updates every 15 minutes while you are awake, wearing the ring and relatively inactive. It needs five days of continuous wear, recalibrates baselines daily, and uses no fixed thresholds. Output is four zones — Stressed, Engaged, Relaxed, Restored. Resilience is a 14-day average built from daytime stress load, daytime recovery and nighttime recovery.

WHOOP — Stress Monitor, 0 to 3. Compares real-time heart rate and HRV against a trailing 14-day baseline. Motion is taken into account to limit — WHOOP's word — the effect of exercise. WHOOP routes psychological stress to a separate self-report Journal, and its own band descriptor for the top of the scale is "stressed or excited." There is no stress endpoint in the public API.

Google, formerly Fitbit. The Fitbit stress score no longer exists. This changed in 2026 and most comparisons have not caught up. On 19 May 2026 the Fitbit app became the Google Health app, and the numeric Stress Management Score was retired. It is replaced by Resilience, reported as Optimal, Balanced or Low, built from more than ten factors across responsiveness (heart rate, HRV and electrodermal activity where available), exertion balance and sleep patterns. The separate Readiness Score now uses three inputs — sleep, HRV and resting heart rate — after Google removed the activity component.

Samsung — stress measurement and Energy Score. Samsung's Gulf support page states the watch uses HRV to estimate stress and that a measurement takes one to two minutes while you stay quiet and still. The scale, the bands and the specific HRV metric are not published, and there is no stress data type in the public Samsung Health SDK. Energy Score was developed with Prof. Patrick O'Connor at the University of Georgia; his own framing is narrow — the score "reflects predicted variation in the ability to perform brief cognitive tests."

Apple — no stress score. This is verifiable three ways: HealthKit's quantity type list contains no stress metric, Apple's Mental Wellbeing feature set is State of Mind logging plus the PHQ-9 and GAD-7 assessments, and the Vitals metric list has none either. Apple offers you a mood slider and overnight vitals against your own baseline, and declines to convert either into a stress score.

Welltory. Spot measurement rather than continuous: smartphone camera, or a chest strap for ECG-grade input. Reports stress, energy, focus and a set of raw HRV metrics — SDNN, RMSSD, pNN50, Mean RR and others — with frequency-domain measures when there are at least 300 intervals. Welltory's own caveat on the raw panel: these are not personalised values, they are based on general population norms.

Side by side

Compiled from each company's own documentation on 22 September 2026. Blank cells mean the company does not publish the figure, which is itself informative.

Two things stand out. Only one company publishes how often the measurement is taken, and only one names the underlying statistics rather than presenting a composite. Everywhere else you are asked to trust a number whose recipe is not disclosed — which is fine for a wellness feature, and worth knowing before you build a habit around the reading.

The scales are also not convertible. A Garmin 60, an Oura "Engaged" and a WHOOP 2 are not three ways of saying the same thing, and moving between devices will not give you a continuous history.

The study that should change how you read all of these

In January 2025 a pre-registered study tested Garmin's Stress Level score against a proper reference. Sixty valid participants wore a Garmin Vívosmart 4 and a Polar H10 chest strap through rest, a standardised mental-arithmetic stressor modelled on the Montreal Imaging Stress Task with a money incentive, and recovery. Participants also completed the PSS-14, the standard perceived-stress questionnaire.

Three findings, and each one matters.

The score tracked heart rate, not HRV. Correlation with mean heart rate was r = 0.84 to 0.85. Correlation with RMSSD — the vagal HRV measure the score is supposed to be built on — was −0.59 to −0.63. Within subjects the gap widened: heart rate r = 0.74, RMSSD r = −0.41.

RMSSD itself barely moved. Under a validated laboratory stressor, mean RMSSD decreased during the stress phase and recovered afterwards, but the difference did not reach significance. The Garmin score rose significantly — because heart rate rose.

The score had no relationship with how stressed people actually felt. Against PSS-14, the Garmin stress score correlated r = −0.17, p = 0.20. Not weakly — not at all. Every other physiological measure was equally null, and so, nearly, was momentary self-report.

This is a preprint and has not been peer-reviewed, which we say plainly. It is also the only pre-registered head-to-head test of a consumer stress score that exists, and its design is sound.

The reasonable conclusion is not that the number is fake. It is that it measures physiological arousal, mostly via heart rate, and that arousal and felt stress are different things that only sometimes coincide.

Does HRV actually measure stress?

The methodologists have been saying this for years, and the vendors' own documents partly agree.

Laborde, Mosley and Thayer's 2017 recommendations in Frontiers in Psychology are the standard reference for doing this properly. On the LF/HF ratio, which much of the commercial lineage rests on: the sympatho-vagal balance interpretation "has been highly criticized," and "there is now a consensus to say that the precise physiological underpinning is unclear." They recommend indices that reflect clearly identified physiological systems — RMSSD and high-frequency HRV — rather than composite ratios.

Firstbeat, whose algorithm sits under Garmin's score, concedes the core problem in its own white paper: "Physiologically the response to positive and negative stress is similar," and there is "an ongoing debate of the exact definition of stress in the scientific literature."

Oura says it more bluntly than anyone: Daytime Stress "cannot differentiate different stress sources, as the body physiologically reacts the same way to different stressors, whether it's physical or mental." And: "Stress is measured based on your biometrics, not your emotions."

WHOOP's own band label — "stressed or excited" — admits the same thing in three words.

The conditions required to measure HRV meaningfully

Laborde and colleagues list what has to be controlled for an HRV measurement to mean something. Stable factors: age, sex, smoking, habitual alcohol, weight and waist-to-hip ratio, cardioactive medication, oral contraceptive use. Transient factors: normal sleep the night before, no intense training the previous day, no meal in the last two hours, no coffee or tea in the last two hours, empty bladder, no alcohol for 24 hours, and a constant time of day.

Now picture a wrist tracker scoring you at three in the afternoon, after a coffee, a sandwich, a phone call and a flight of stairs.

That is not an argument against wearing one. It is the reason a single reading means little and a pattern across weeks means more — and the reason a controlled spot measurement, taken the same way each morning, is a genuinely different kind of data from a number generated continuously in the middle of life.

What makes a stress score go up besides stress?

• Caffeine — Oura, Google, Firstbeat white paper

• Alcohol — Apple (Vitals), Firstbeat

• Illness — Apple, Google, Firstbeat

• Posture — Firstbeat, Google, Samsung, Welltory

• Movement — Garmin (grey bars), Oura, WHOOP, Google

• Talking — Welltory

• Breathing rate — Firstbeat, Welltory

• Medication, cardiac conditions — Firstbeat, Oura, Apple, Welltory

One contradiction is worth naming. Garmin's manual states that food intake and stimulants like caffeine have no impact on Body Battery. Firstbeat's own white paper — Firstbeat being a Garmin subsidiary and the source of the algorithm — lists stimulants such as coffee among the factors affecting the stress measurement.

Validation, vendor by vendor

None of these scores has been validated against an accepted reference in a peer-reviewed study of that score. Here is what each company offers instead.

Firstbeat reports correlations of r = 0.40–0.50 between cortisol after waking and its sleep-time stress indicators — from a 2006 conference abstract, not a journal paper. Its peer-reviewed support is Uusitalo and colleagues in Applied Ergonomics 2011, relating work effort to daytime HRV, r = −0.66 and −0.54.

Oura states that Daytime Stress "has been validated by conducting internal studies where we compared Oura results to other well-established industry solutions." No comparator, sample size or date is named, and Oura's own Science and Research page lists validations for sleep staging, heart rate, ovulation and blood oxygen — not for stress.

WHOOP promotes a 2025 JMIR paper of its own authorship: over 300,000 monthly mental-health surveys from more than 170,000 members against 7.9 million days of biometrics, correlating HRV, resting heart rate and sleep consistency with PSS, GAD-2 and PHQ-2. It is a real and interesting study. It does not test the Stress Monitor output, so it is not a validation of the score.

Google publishes no validation of Resilience or the retired Stress Management Score.

Welltory. Our peer-reviewed paper validates signal acquisition — the camera against ECG — not the stress metric. The only stress-specific study is a 2020 medRxiv preprint of 1,471 adults comparing HRV-derived stress with perceived stress. A preprint is not peer review, and we are not going to describe our stress metric as validated.

Regulatory status is uniform. No stress feature from Garmin, Google, Samsung, Oura, WHOOP or Welltory holds FDA clearance or a CE mark as a medical device. Apple has twelve FDA device records and none of them is a stress feature, because it has none. WHOOP's 2025 clearance is frequently reported as covering stress; it covers Heart Screener, an on-demand ECG on WHOOP MG.

So how should you read a stress number?

Treat HRV stress measurement as arousal, not emotion. A high reading means your cardiovascular system is working harder relative to your own baseline. Coffee does that. Stairs do that. A tense call does that. Excitement does that. Coming down with something does that.

Read the pattern, not the point. One afternoon tells you almost nothing. Four weeks of afternoons, compared against what you know you were doing, occasionally tells you something you did not know — usually about sleep debt, alcohol, or a week that was heavier than it felt.

Prefer the measurement you control. If you take a reading the same way each morning, before coffee, sitting still, you remove most of the confounds in the Laborde list at once. That is the single biggest accuracy improvement available to anyone in this category, and it costs nothing.

And notice which companies tell you what they cannot do. Oura publishes its sampling interval and states its score cannot separate physical from mental stress. Firstbeat states that positive and negative stress look the same physiologically. Both are statements about the limits of the measurement, published by the companies themselves.

What none of this can do

No stress score diagnoses anything. Anxiety disorders, depression, burnout and thyroid disease are clinical diagnoses made by clinicians, and no wearable claims otherwise.

No stress score predicts anything. None of these companies claims a validated predictive function, and none of the evidence supports one.

No stress score distinguishes causes. The same rise appears for a deadline, a double espresso, a fever and good news.

How to bring this up with your doctor

Bring the raw numbers, not the score. Resting heart rate over weeks is the most interpretable output a consumer device produces, and the one a clinician is most likely to engage with. "My stress is 78" means nothing outside the app that generated it.

Name the statistic and the conditions. "My RMSSD measured sitting, before coffee, has fallen from about 45 to about 30 over six weeks" is a sentence a doctor can use. If a change persists and you cannot explain it, that is the moment for a conversation rather than a settings change.

How Welltory fits

Welltory reports stress from the same physiological basis as everyone here, and names the statistic it computes rather than presenting an undisclosed composite. It reads Apple Watch, Samsung Watch and Bluetooth heart-rate monitors, and pulls data from Apple Health, Health Connect, Samsung Health, Oura and Garmin — so it can read your existing device rather than replace it.

Like every stress feature in this comparison, Welltory's holds no regulatory clearance, and its only dedicated stress study is a preprint rather than peer-reviewed work. Its peer-reviewed validation, 26 participants against simultaneous ECG, covers signal acquisition.

What that leaves is the one thing this article keeps returning to. The pre-registered test found a continuous score tracking heart rate rather than stress, because continuous scoring cannot tell a deadline from a flight of stairs. A reading you take deliberately — same time, sitting, before coffee — removes most of the conditions the research says have to be controlled, and it is the single accuracy improvement available to anyone here at no cost.

How we made it

Claims about other companies come from their own support pages, developer documentation, manuals, white papers and FDA records, retrieved on 22 September 2026, and from named studies. Where a company's own sources disagree with each other, both are reported. Where a figure could not be found in a primary source, it was left out rather than filled in.

Made with AI tools, then edited and fact-checked by people.

Data analysis by Jane Smorodnikova — the founder and CEO of Welltory. A recognized tech leader with two Master's degrees and experience at MIT, she has scaled Welltory to over 17 million users.

Written by Tatsiana Yashyna — Deputy COO at Welltory. With a background in medicine and years of working with health data, she translates research and real physiological signals — sleep, stress, heart rate, and hormones — into clear, evidence-based explanations that help people understand what their bodies are telling them.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data