Health & Fitness GearHow-to
How Accurate Are Fitness Trackers? What the Research Shows
How accurate are fitness trackers? Peer-reviewed studies on heart rate, steps, sleep, VO2 max and calories, and why calorie counts are the least reliable.
By PlusReviews Editorial TeamPublished 11 min read12 sources
Disclosure: PlusReviews is reader-supported. When you buy through links on this page we may earn an affiliate commission, at no extra cost to you. Our picks are based on independent research. How we make money.
Fitness trackers are accurate enough for trends, but not for exact numbers, and how accurate depends on what you are measuring. Peer-reviewed studies find that wrist trackers are good at heart rate at rest and during steady cardio, and reasonable at counting steps. They are weaker at sleep stages and VO2 max, and they are worst at calories burned, where errors of 20% or more for an individual are common. If you remember one rule, make it this: compare your tracker with itself over weeks, not with a lab test on a single day.
Accuracy matters when you are choosing a device. Our guide to the best fitness trackers ranks bands, straps, watches and rings on comfort, battery and cost, and this guide explains how far you can trust the numbers they all produce. If you are deciding between device types, fitness tracker vs smartwatch covers that, and a bigger, pricier device does not automatically mean better data.
This guide is research-based. We read the peer-reviewed validation studies and reviews listed in the sources, most of which compare consumer devices with lab-grade reference equipment. We did not run our own tests. One caveat applies throughout: published studies lag behind product launches, so most tested older models, such as the Fitbit Charge 2 and the Apple Watch Series 6 to 8. They show how the technology behaves, not how a specific 2026 model performs.
| Metric | How it is measured | What studies found | How to use it |
|---|---|---|---|
| Heart rate | Optical sensor (PPG) on the skin | Average bias about 3%; worse during fast changes and low wrist motion | Trust resting and steady-state readings; use a chest strap for intervals |
| Steps | Accelerometer and step algorithm | Errors from about 9% under to 12% over | Watch weekly averages, not single days |
| Sleep | Movement plus heart rate | Detects sleep well, misses time awake; overstates total sleep by 10% or more | Track bedtime consistency and long-term trends |
| VO2 max | Model built on heart rate, pace and profile | Overestimates by about 10 to 15% in pooled studies | Use as a fitness trend line, not a lab result |
| Calories | Model built on heart rate, motion and profile | Large individual errors; one study found no device under 20% | Compare with yourself only; never eat back the number |
Heart rate: the most reliable number your tracker gives you
Wrist trackers measure heart rate with photoplethysmography (PPG). Green LEDs shine into the skin, and a sensor reads tiny changes in reflected light as blood pulses through the vessels. It is an indirect method, but it works well when the sensor sits still against the skin.
The best summary of the evidence is a 2024 “living” umbrella review in Sports Medicine by researchers at University College Dublin. It pooled 24 systematic reviews covering 249 validation studies and found that wearables measured heart rate with an average bias of about 3%. Older individual studies point the same way. A 2017 Stanford study in the Journal of Personalized Medicine tested seven wrist devices on 60 adults and found that six of them had a median heart rate error below 5% during cycling.
The weak spots are just as consistent across studies:
- Fast changes and intervals. A 2020 Duke University study in npj Digital Medicine found that absolute heart rate error during activity was, on average, about 30% higher than at rest. Trackers keep up with a steady pace better than with sudden sprints and recoveries.
- Low wrist motion. In a 2018 study in JMIR mHealth and uHealth, a Fitbit Charge 2 and a Garmin vivosmart HR+ were most accurate during treadmill intervals and least accurate during high-intensity bike intervals, where they underestimated heart rate by about 11% and 14% on average. Accuracy also dropped when participants did not start a workout mode on the device.
- Weightlifting. Gripping a bar or dumbbell flexes the forearm and moves the sensor. A study of the Apple Watch 6, Polar Vantage V and Fitbit Sense, published in the European Journal of Sport Science in 2022, found the Apple Watch held a coefficient of variation under 5% across all five activities, including resistance exercise, while the Polar and Fitbit ranged from high to poor accuracy depending on the activity.
What about skin tone? The evidence is mixed. The 2017 Stanford study found more error for people with darker skin, as well as for men and for people with a higher BMI. The 2020 Duke study, which was designed specifically to cover the full range of skin tones, found no statistically significant difference in accuracy by skin tone, but significant differences between devices and between activities. The fair reading is that motion and device design matter more than skin tone in recent testing, but the question is not closed.
Steps: good enough for daily targets
Step counting uses an accelerometer and an algorithm that decides which movements are steps. The 2024 Sports Medicine umbrella review found mean absolute percentage errors ranging from about 9% under to 12% over, with wearables more often undercounting. A 2020 systematic review of Polar trackers in the International Journal of Exercise Science reached a similar conclusion: step counts showed less average error than calorie or intensity estimates, though study methods varied widely.
In practice, step counts go wrong in predictable situations. Slow, shuffling walking may not register. Pushing a stroller or shopping cart keeps the wrist still, so steps can be missed. Arm-heavy activities, such as folding laundry or driving on a bumpy road, can add phantom steps. None of that matters much if you look at weekly averages, because the same biases repeat. A jump from 6,000 to 8,000 steps a day is real even if neither number is exact.
Sleep: great at “asleep,” weak at “awake”
Consumer trackers guess sleep from movement and heart rate, while the reference standard, polysomnography, records brain waves, eye movements and muscle activity in a sleep lab. A tracker is therefore inferring sleep, not observing it.
A 2021 study in the journal Sleep, run by the US Naval Health Research Center, tested seven consumer sleep trackers against polysomnography over three nights. Every device was very good at detecting sleep, with a sensitivity of 0.93 or higher, but much weaker at detecting wake, with a specificity of only 0.18 to 0.54. In plain terms, if you lie still but awake, the tracker often counts it as sleep. Devices also did worse on nights when sleep was deliberately disrupted, and their sleep-stage results were inconsistent. That pattern explains the umbrella review’s finding that wearables tend to overestimate total sleep time, typically with errors above 10%.
Sleep stages are the shakiest part. A 2023 multicenter study in JMIR mHealth and uHealth compared 11 consumer sleep trackers, including the Apple Watch 8, Fitbit Sense 2, Google Pixel Watch and Oura Ring 3, with polysomnography in 75 participants. Agreement on sleep stages varied widely, with scores from 0.26 to 0.69 on a 0-to-1 scale. Several of its authors are employees of Asleep Co., a sleep-tracking company, so read its ranking of individual devices with that in mind.
Sleep trackers are still useful for habits: bedtime consistency, total time in bed and how your nights trend after you change caffeine, alcohol or screen time. If your goal is better sleep, the comfort of the device matters more than small accuracy differences. That is why the Oura Ring 4 is the sleep pick in our best fitness trackers guide.
VO2 max: a useful trend line, not a lab result
VO2 max is the most oxygen your body can use during hard exercise, and it is a common measure of cardiorespiratory fitness. A lab test means running or cycling to exhaustion while wearing a mask that analyzes your breath. A watch instead estimates it from your heart rate, pace or power and your profile.
The 2024 umbrella review found that wearables overestimated VO2 max by about 15% when the estimate relied on resting data and by about 10% when it used exercise data. A 2022 meta-analysis by the European INTERLIVE network in Sports Medicine reached the same conclusion: estimates based on workouts were more accurate than those based on resting conditions, but the error for an individual was still large, so the authors said the methods need improvement for sports and clinical use.
A 2024 study of the Apple Watch Series 7 in JMIR Biomedical Engineering is a good example. With 19 participants, the watch’s average estimate was lower than the lab value, with a mean absolute percentage error of about 16%. The authors noted that the watch tended to overestimate VO2 max in people with low fitness and underestimate it in very fit people. So if your watch says your VO2 max went up after three months of training, that trend is probably real. The exact figure is less trustworthy.
Calories: why they are the least accurate number on your wrist
Calories burned is the metric people care about most and the one trackers get most wrong. The reason is simple: a tracker never measures calories. It measures heart rate and movement, then feeds those numbers, plus your age, sex, height and weight, into a company’s own model to estimate energy use. Every step in that chain adds error.
- The inputs are imperfect. Any heart rate or motion error flows straight into the calorie estimate.
- The models are built on averages. Two people with the same heart rate and weight can burn quite different amounts of energy, depending on fitness, body composition and how efficiently they move. A model built on population averages cannot see those differences.
- Heart rate rises for reasons other than effort. Heat, stress, caffeine and dehydration all raise heart rate without burning many extra calories.
- Wrist motion does not match whole-body work. Cycling, rowing, carrying loads and lifting weights involve a lot of effort with little or unusual arm movement.
- Errors add up over a day. A 20% error on a 300-calorie workout is 60 calories. Spread a similar error across a full day’s total estimate, which also includes a modeled resting burn, and the gap can run into the hundreds.
Lab studies confirm it. In the 2017 Stanford study, which used indirect calorimetry (measuring oxygen in and carbon dioxide out) as the reference, no device achieved a calorie error below 20%. The 2022 European Journal of Sport Science study found poor calorie accuracy for all three watches in all five activities, with coefficients of variation between about 13% and 30%. A 2020 meta-analysis of 60 studies in the British Journal of Sports Medicine found that accuracy varied widely by activity type, and that combining heart rate with motion sensing reduced the error in most activities.
Averages can hide the problem. The 2024 umbrella review reported a small average underestimate of about 3%, but the errors across the pooled studies ranged from about 21% under to 15% over, and single-device lab studies show far wider spreads between individual people. A device can look fine on average across a study group and still be far off for you.
The newest evidence adds a twist. A study from Florida International University, published in PLOS One in July 2026, tested an Apple Watch Series 8, Fitbit Sense 2, Samsung Galaxy Watch 5 and Garmin Forerunner 955 on 58 Hispanic adults during a 10-minute interval session on a recumbent bike. Three of the four overestimated calories on average: the Garmin by about 69 calories, the Samsung by about 57 and the Apple Watch by about 22. The Fitbit’s average bias was near zero, but the authors note that this result depended on removing its outlier readings. Error also rose with body fat across all four brands. That is a short, single-activity test, but it points to the same conclusion as everything before it: calorie numbers are not reliable enough to plan meals around.
How to get more accurate readings from any tracker
You cannot fix a company’s algorithm, but you can remove the most common sources of error:
- Wear it snugly, a finger’s width above the wrist bone. A loose band lets light leak in and the sensor slide. Tighten it a notch for workouts and loosen it afterward.
- Start a workout mode. Trackers sample heart rate more often and use activity-specific models when they know you are exercising. The 2018 JMIR study found accuracy dropped when participants skipped this.
- Keep your profile up to date. Age, height and especially weight feed the calorie and VO2 max models.
- Pair a chest strap for intervals and lifting. Many Garmin, Polar, Apple and Wear OS devices accept a Bluetooth heart rate strap, which measures the heart’s electrical signal directly. Check your device’s compatibility first.
- Compare like with like. Switching brands resets your baseline, because each company uses different models. Our Fitbit alternatives guide notes that changing brand rarely fixes calorie accuracy.
- Look at weekly averages. Day-to-day noise mostly cancels out over a week.
If your goal is strength, a tracker’s calorie count is the least useful number it produces. Progressive training with simple equipment, such as the adjustable dumbbells in our home gym guides, will do more for your results than a more precise estimate.
How we researched
We started from the 2024 living umbrella review in Sports Medicine, which summarizes 24 systematic reviews, then read the individual studies that best illustrate each metric. For heart rate and calories, we focused on lab studies that compared devices with chest-strap ECG and indirect calorimetry. For sleep, we drew on studies that compared trackers with polysomnography, and for VO2 max, a meta-analysis and a single-device validation against lab gas analysis. We noted each study’s size, date, devices and funding where relevant, and we avoided company-run validation claims. Because most studies tested devices that are now one or more generations old, we describe what the technology does well and badly rather than ranking current models.
The bottom line
Fitness trackers are accurate enough to be useful and inaccurate enough to mislead you if you read them literally. Trust resting heart rate and steady-state workout heart rate the most, steps and sleep timing next, and VO2 max as a long-term trend. Treat calories as the least reliable estimate on the device. Wear the tracker snugly, start workout modes, and judge progress over weeks. For help choosing a device, see our best fitness trackers guide.
Frequently asked questions
How accurate are fitness trackers for heart rate?
Fairly accurate at rest and during steady cardio. A 2024 umbrella review in Sports Medicine, covering 24 systematic reviews, found an average heart rate bias of about 3%. Accuracy drops during intervals, weightlifting and high-intensity bike intervals, when heart rate changes fast or the wrist barely moves.
Are fitness tracker step counts accurate?
Close enough for daily trends. The same 2024 Sports Medicine review found step-count errors ranging from about 9% under to 12% over, with devices more often undercounting. Slow walking, pushing a stroller or a shopping cart, and wearing the device loosely all make step counts less reliable.
Why are fitness tracker calorie counts so inaccurate?
Because a tracker does not measure calories at all. It estimates them from your heart rate, movement and profile details such as age and weight, using a company's own formula. Lab studies against breathing-gas analyzers have repeatedly found large errors for individuals, and a 2026 study in PLOS One found the error grew with body fat.
Can a fitness tracker measure sleep accurately?
It is good at telling when you are asleep and weaker at spotting when you are awake in bed, so it tends to overstate total sleep. Sleep-stage breakdowns, such as deep and REM sleep, vary a lot between devices compared with a lab sleep study. Use the trends, not single nights.
Is the VO2 max on my watch accurate?
It is a reasonable estimate across groups of people but can be well off for one person. A 2024 study of the Apple Watch Series 7 found an average error of about 16% against a lab test, and reviews find estimates based on workouts are more accurate than those based on resting data alone.
Sources
The specifications, test results and documentation this guide draws on:
- Keeping Pace with Wearables: A Living Umbrella Review (Sports Medicine) — Doherty et al., Sports Med 54(11), November 2024; 24 systematic reviews, 249 validation studies; heart rate, steps, calories, VO2 max and sleep summary figures
- Accuracy in Wrist-Worn, Sensor-Based Measurements of Heart Rate and Energy Expenditure in a Diverse Cohort (Journal of Personalized Medicine) — Shcherbina et al., Stanford, May 2017; 7 devices, 60 adults; no device had calorie error below 20%
- Investigating sources of inaccuracy in wearable optical heart rate sensors (npj Digital Medicine) — Bent et al., Duke, February 2020; no significant accuracy difference across skin tones; error about 30% higher during activity than rest
- Accuracy of Wrist-Worn Activity Monitors During Common Daily Physical Activities and Types of Structured Exercise (JMIR mHealth and uHealth) — Reddy et al., December 2018; Fitbit Charge 2 and Garmin vivosmart HR+; largest heart rate error during bike intervals
- Wrist-worn devices for the measurement of heart rate and energy expenditure (European Journal of Sport Science) — Hajj-Boutros et al., published online January 31, 2022 (February 2023 issue); Apple Watch 6, Polar Vantage V, Fitbit Sense across five activities including resistance exercise
- How well do activity monitors estimate energy expenditure? (British Journal of Sports Medicine) — O'Driscoll et al., March 2020 issue; meta-analysis of 60 studies; accuracy varies by activity, heart rate plus accelerometry improves estimates
- Measuring Physical Activity Using Triaxial Wrist Worn Polar Activity Trackers: A Systematic Review (International Journal of Exercise Science) — Henriksen et al., May 2020; 14 studies; step counts showed less average error than energy expenditure
- Body fat, skin tone, and the accuracy of smartwatch caloric expenditure estimates (PLOS One) — Kostrna et al., Florida International University, July 29, 2026; 58 adults, 10-minute cycling protocol, Apple, Fitbit, Samsung and Garmin bias figures
- Performance of seven consumer sleep-tracking devices compared with polysomnography (Sleep) — Chinoy et al., Naval Health Research Center, May 2021; sensitivity 0.93 or higher, specificity 0.18 to 0.54
- Accuracy of 11 Wearable, Nearable, and Airable Consumer Sleep Trackers (JMIR mHealth and uHealth) — Lee et al., November 2023; 75 participants; sleep-stage F1 scores from 0.26 to 0.69; several authors are employees of Asleep Co.
- Validity of Estimating the Maximal Oxygen Consumption by Consumer Wearables (Sports Medicine) — Molina-Garcia et al., INTERLIVE network, July 2022; exercise-based VO2 max estimates more accurate than resting-based ones
- Assessing the Accuracy of Smartwatch-Based Estimation of Maximum Oxygen Uptake Using the Apple Watch Series 7 (JMIR Biomedical Engineering) — Caserman et al., July 31, 2024; 19 participants; mean absolute percentage error 15.79%