Fitness trackers and smartwatches report steps, calories and activity levels with confident precision, and many people use those numbers to judge their training or compare equipment. Whether the numbers are accurate is a research question. This guide summarises two published reviews: a 2020 systematic review by Fuller and colleagues in JMIR mHealth and uHealth, and a 2024 “living” umbrella review by Doherty and colleagues in Sports Medicine. It concerns consumer wearables only; it does not assess the built-in consoles on treadmills, bikes or rowers, which are covered in our guide to console metrics and their accuracy. It is general information, not medical advice.
The 2020 systematic review
Fuller and colleagues searched PubMed, Embase and SPORTDiscus for English-language articles up to May 2019 and included 158 publications examining nine commercial wearable brands, with Fitbit by far the most studied. In laboratory settings, Fitbit, Apple Watch and Samsung appeared to measure steps accurately. Heart rate results were more variable, with Apple Watch and Garmin the most accurate and Fitbit tending to underestimate. For energy expenditure, the authors report that no brand was accurate. They conclude that wearables are accurate for steps and heart rate in laboratory-based settings, but that this varies by manufacturer and device type, and they note that devices are constantly being redesigned, so more current reviews are needed.
The 2024 umbrella review
Doherty and colleagues reviewed the reviews. From 904 records they included 24 systematic reviews or meta-analyses, covering 249 non-duplicate validation studies and 430,465 participants (43% female), and extracted measures such as mean absolute percentage error and pooled bias. Their headline findings on the outcomes most relevant to home exercisers were:
- Steps: wearables mostly underestimated step counts, with mean absolute percentage errors ranging from −9% to 12%.
- Energy expenditure: mean bias of about −3 kcal per minute, or −3%, with error ranging from −21.27% to 14.76%.
- Heart rate: a mean bias of ±3%.
- Physical activity intensity: mean absolute error ranging from 29% to 80%, depending on the intensity of the activity.
- Aerobic capacity: VO2max significantly overestimated, by ±15.24% during resting tests and ±9.83% during exercise tests.
What the two reviews say together
Both point the same way on the measurements that matter to most users. Steps are the best-measured outcome in laboratory conditions, with the umbrella review reporting a tendency towards underestimation. Calorie estimates are the weakest: the 2020 review found no brand accurate for energy expenditure, and the 2024 review reports wide error ranges even though the average bias was small. A small average bias can hide large errors in individual readings, which is why the range matters more than the mean for someone using a number to plan meals or training. The 2024 review also reports that intensity measurement had large errors, so “active minutes” and zone targets deserve caution.
What the evidence cannot say
The umbrella review states that approximately 11% of commercially available wearable devices released to date have been validated for at least one biometric outcome, and that the number of validation studies represents just 3.5% of the total needed for a comprehensive evaluation. Its authors say that a conclusive assessment is impeded by heterogeneity in outcomes and methods, and call for standardised validation protocols. The 2020 review is based on data up to May 2019, so newer models are not covered. Neither review supports a ranking of brands for current products, and neither addresses machine consoles. The 2020 conclusions are specific to laboratory-based settings, and the review defined validity as agreement with other measures (criterion validity) and with what the device claims to measure (construct validity), with reliability assessed within and between devices. It included studies of the general population and of special populations, so its results are not restricted to one group of users.
Using trackers sensibly
Trends are more useful than single readings: a device that consistently underestimates steps still shows whether a week was busier than the last, provided the same device is used. Heart rate readings are covered in more detail in our guide to chest straps against optical sensors, and the wider marketing pattern in fitness gadgets and gimmicks.
Frequently asked questions
Are calorie counts on trackers reliable? The 2020 review found no brand accurate for energy expenditure, and the 2024 review reports errors ranging from −21.27% to 14.76%.
Do trackers over- or under-count steps? The umbrella review found that wearables mostly underestimated step counts.
Does brand matter? The 2020 review reports that accuracy varied by manufacturer and device type and that devices change quickly.
The bottom line
Reviews of wearable trackers show that step counts and heart rate are reasonably accurate in laboratory conditions, that steps tend to be underestimated, and that calorie and intensity estimates carry wide errors. The numbers are best used to follow personal trends rather than as precise measurements.
Sources
- Fuller D et al., “Reliability and Validity of Commercially Available Wearable Devices for Measuring Steps, Energy Expenditure, and Heart Rate: Systematic Review,” JMIR mHealth and uHealth, 2020
- Doherty C et al., “Keeping Pace with Wearables: A Living Umbrella Review of Systematic Reviews Evaluating the Accuracy of Consumer Wearable Technologies in Health Measurement,” Sports Medicine, 2024