Steps and calories on your watch: why they're not equally reliable
AI-generated content. How we make it
In brief
- Counting steps means recognizing a pattern that movement leaves in the sensor; calories aren't measured, they're calculated with a model. In the lab, almost half of the step comparisons fell within a 3% margin; of the calorie comparisons, fewer than one in ten.1
- The watch calculates calories from movement, heart rate and details such as your height and weight, using formulas that manufacturers don't usually publish.123 In studies, the average error of that figure is around 30% or more, and in some models it's much larger.53
- Steps also go wrong in specific situations: when walking slowly, pushing a stroller or a cart, walking with a walker or moving your arms without walking.879
- As an inference from the size of those errors: steps are useful for comparing your weeks with the same watch; calories, at most, as a rough guide, not for deciding how much to eat.
At the end of the day, your watch gives you two figures that look equally precise: so many steps, so many calories. Behind each one is a very different job, and the studies that compare these devices with reference methods show it quite clearly.
This article explains where each figure comes from, why one is much closer to reality than the other, in which situations even the step counter fails and what reasonable use you can make of each.
Two figures obtained in very different ways
To count steps, the watch uses an accelerometer. Each stride leaves a peak and a trough in its signal that repeat at a regular rhythm, and the algorithm counts those cycles.7 It's a count: there's something concrete to count, and it can be checked by counting steps by hand or on video, which is how these devices are validated.14
With calories, it's a different story. No sensor in the watch measures the energy you burn. In a lab, the reference is to measure the oxygen you take in and the carbon dioxide you breathe out or, over several days, a method using water labeled with isotopes.1 The watch, by contrast, estimates that energy from movement and heart rate, which it combines with personal details such as height and weight through each manufacturer's own formulas.32 According to the authors of one review, only one of the brands studied used a published equation for resting expenditure; the others, as far as the authors knew, didn't publish how they calculated calories.1
The same movement doesn't cost everyone the same energy. The authors of a validation study note, citing a handbook of metabolic calculations, that 10,000 steps can amount to anywhere between 400 and 800 kilocalories, depending on the person's height and weight.2 A model has to guess those differences; a step counter doesn't.
Figure 1. In the lab, steps and heart rate are on target far more often than calories
Why calories are off even when heart rate is right
In a study with 60 volunteers of varied ages, sizes and skin tones, six of the seven watches and bands tested measured heart rate with a median error below 5% while cycling. For calories, none came close: the median error was 27.4% for the best and 92.6% for the worst. The error was larger when sitting (52.4%) than when walking or running (around 31%).2
A study published in 2026 tested four watches from well-known brands, 2022 models, in 58 Hispanic adults aged 18 to 50 (with an average age of 23) during ten minutes on a stationary bike. On average, each watch was off on exercise calories by between 32% and 78%, depending on the model, and three of the four tended to overestimate them. The error grew with body fat percentage for all of them; skin tone, within the range studied, showed no clear effect.3
Why they're so far off isn't entirely clear. The authors of the first study point out that the energy cost of the same activity varies a lot from one person to another, and that not all algorithms make use of heart rate.2 Those of the second note that many models have been calibrated on leaner people.3 On heart rate, the reviews don't agree: a meta-analysis of 60 studies found that combining it with movement reduced the error in most activities,6 and another review saw no clear relationship between having a heart-rate sensor and being more accurate.1
When the step counter gets it most wrong
The step counter works best when you walk at a normal or brisk pace, with your arms swinging naturally. The further your movement strays from that pattern, the harder it is for the counter.78
| Situation | What they found | Who took part |
|---|---|---|
| Walking slowly | The ten devices in one study generally performed worse at 3.2 km/h (2 mph) than at 4.8 or 6.4 km/h (3 or 4 mph)9; in another, with three wristbands, accuracy improved as walking speed increased8 | 31 and 30 healthy adults, on a treadmill |
| Pushing a stroller or a shopping cart | With a stroller, the three wristbands in one study missed more than a third of the steps8; with a cart, the wristband in another study had a mean absolute error of 19.8%, compared with 6.4% for the device carried in a pants pocket, although its authors found no systematic deviation from the actual count7 | 30 and 15 young, healthy adults |
| Walking with a walker | The wristband was off by 73.1% of the steps7 | 15 young, healthy adults who used a walker for the test |
| Moving your arms without walking | The wristbands added steps while folding laundry8 and about 4.8 steps per minute while eating seated7 | 30 and 15 young, healthy adults |
| Wrist or hip | Compared with a hip pedometer, a wristband counted 6% fewer steps on a treadmill and about 7,695 steps against 6,309 in everyday activities, around 22% more10 | 86 adults aged 18 to 65 |
| Dominant wrist | In everyday activities, the error was larger with the wristband on the dominant hand8 | 30 young, healthy adults |
A practical consequence, inferred from these results: if you walk slowly, often push a stroller or use a walker, the figure may come in systematically low, and a day with lots of arm-based chores may add steps you didn't take. What happens to step error outside the lab, in daily life, is summarized in the table of the article on wearables.
What each figure is useful for
Steps, for comparing yourself with yourself. If you wear the same watch on the same wrist and your routine is similar from one week to the next, the error tends to be similar too, and a change in the weekly average tells you something. How many days it takes for that average to be reliable we explain in why a single data point is misleading. If one week you change your activity—more walks with a stroller, more housework—the figure can move without you having walked more or less.
Calories, at most, as a rough guide. With average errors of 30% or more in the studies, they shouldn't be used to decide how much to eat or to “make up for” what you've burned.53 The authors of the 2026 study leave one possibility open: that the figure, even if it's not accurate in absolute terms, could be used to track changes in the same person, something that hasn't yet been tested in daily life.3
If what you want is to move more, you don't need an exact figure: any amount of activity counts, and steps or minutes walking are enough to see whether you're getting there. If you want to change your diet or your weight for health reasons, a health professional, such as a dietitian, will help you more than your watch's calorie counter.
How many steps a day are associated with better health is a different question, with different studies. Where the 10,000 figure comes from and what studies show about steps we cover in how many steps a day.
Keep reading
Wearables: useful data vs. noise
How reliable each piece of data from a watch or a ring is, from heart rate to sleep, and how to look at it without turning it into a daily judge.
How to turn a wellness data point into a question and a test
If your step average changes and you want to know why: how to tell normal variation from a change and test a habit without fooling yourself.
Healthy habits when you're short on time: the minimums that really count
How much movement really counts when you're short on time, with no need to count calories.
Watches and rings that track sleep: what they get right and what they don't
The same question applied to sleep: what a watch or a ring can and can't estimate about your nights.
Want to turn information into action?
naro helps you prioritize 2–3 high-potential habits and sustain them over time, with context and follow-up.
Sources
Each small number in the text leads to the source that statement rests on. Practical advice without a number is a deduction from these sources, and examples with a person's name are made up. It is general information: it does not replace an assessment by a health professional. How we make this content
Sources consulted up to 26 October 2026.
- Systematic review. Fuller D, Colwell E, Low J, et al. (2020). Reliability and validity of commercially available wearable devices for measuring steps, energy expenditure, and heart rate: systematic review. JMIR mHealth and uHealth. 158 publications up to May 2019, mostly lab-based; many of the devices are no longer sold.
- Validation study. Shcherbina A, Mattsson CM, Waggott D, et al. (2017). Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. Journal of Personalized Medicine. 60 volunteers, in the lab; seven wrist-worn devices available in 2016.
- Validation study. Kostrna J, Oparina E, Palacios C, et al. (2026). Body fat, skin tone, and the accuracy of smartwatch caloric expenditure estimates. PLOS One. 58 Hispanic adults aged 18 to 50, ten minutes on a stationary bike; four watches.
- Systematic review and meta-analysis. Chevance G, Golaszewski NM, Tipton E, et al. (2022). Accuracy and precision of energy expenditure, heart rate, and steps measured by combined-sensing Fitbits against reference measures: systematic review and meta-analysis. JMIR mHealth and uHealth. 52 studies of a single brand, with mostly young, healthy participants.
- Systematic review. Germini F, Noronha N, Borg Debono V, et al. (2022). Accuracy and acceptability of wrist-wearable activity-tracking devices: systematic review of the literature. Journal of Medical Internet Research. 65 studies up to May 2019, very different from one another; it doesn't pool their results into a single figure.
- Systematic review and meta-analysis. O'Driscoll R, Turicchi J, Beaulieu K, et al. (2020). How well do activity monitors estimate energy expenditure? A systematic review and meta-analysis of the validity of current technologies. British Journal of Sports Medicine. 60 studies with healthy adults and wrist- or arm-worn devices; very heterogeneous results. We could only read the abstract.
- Validation study. Alinia P, Cain C, Fallahzadeh R, et al. (2017). How accurate is your activity tracker? A comparative study of step counts in low-intensity physical activities. JMIR mHealth and uHealth. 15 young, healthy adults; three models from one brand worn on the wrist, the chest and in a pocket.
- Validation study. Chen MD, Kuo CC, Pellegrini CA, Hsu MJ (2016). Accuracy of wristband activity monitors during ambulation and activities. Medicine & Science in Sports & Exercise. 30 young, healthy adults, three wristbands on both wrists. We could only read the abstract.
- Validation study. Fokkema T, Kooiman TJ, Krijnen WP, et al. (2017). Reliability and validity of ten consumer activity trackers depend on walking speed. Medicine & Science in Sports & Exercise. 31 healthy adults, ten devices, three speeds on a treadmill. We could only read the abstract.
- Comparison study. Nelson RK, Hasanaj K, Connolly G, et al. (2022). Comparison of wrist- and hip-worn activity monitors when meeting step guidelines. Preventing Chronic Disease. 86 adults; a wristband compared with a hip pedometer, which can also be wrong.