Accelerometers and Motion Sensing
How the motion sensors in wearables and phones turn movement into step counts, what affects their accuracy, and why steps are measured better than calories burned.
Content last updated . This page has not been medically reviewed.
What the sensor does
An accelerometer is a small motion sensor. Phones, fitness bands and watches use it, sometimes together with a gyroscope (a rotation sensor), to estimate steps from the pattern of movement. [1] The device does not "see" a step. It records movement signals, and software decides which pieces of the signal look like steps. That software is where much of the error comes from.Direct, estimated and calculated
| Output | What it really is |
|---|---|
| Raw movement signal | Measured directly by the sensor |
| Step count | Estimated by an algorithm from the signal |
| Distance, pace, calories burned | Calculated from the estimated steps plus personal details and assumptions |
| Activity or "exercise minutes" scores | Brand-specific scores; no universal clinical meaning |
What studies show for steps
- A systematic review of 158 publications found that Fitbit, Apple Watch and Samsung devices appeared to measure steps accurately in laboratory settings, with variation by manufacturer and device type. [2] Label: established for laboratory walking, with differences between models.
- A meta-analysis of recent Fitbit models found an average underestimate of about 3 steps per minute, which the authors judged acceptable on average for steps. [3]
- An Apple Watch meta-analysis of 56 studies found accuracy for steps was generally acceptable, but some subgroups exceeded a 10% error threshold. [4]
- Calories burned is the weak spot: the first review found no brand accurate for energy expenditure. [2] Treat calorie numbers from a wrist as rough guesses. Label: no good evidence of accuracy.
What can throw steps off
- Where the device is carried. Phone step counting is harder when the phone can sit in a pocket, hand or bag, which is why researchers build algorithms that first work out the carrying position. [1]
- Slow or unusual walking. In a small study of healthy adults simulating walking with crutches, frames and walkers, most commercial wearables counted steps accurately only under specific conditions; a research-grade accelerometer worn at the ankle did best. [5] This was a case series of 20 healthy volunteers, not patients.
- Age, body size and sex. Across a large body of validation studies, step-count error differed for older adults compared with adults, and error was greater in studies of people with overweight. [6]
- Children with muscle disease. In 15 children with Duchenne muscular dystrophy and 15 typically developing children, a waist-worn phone accelerometer with a custom calibration counted steps with a mean error of about 1.5%. [7] That is a research method, not what a consumer band does out of the box.
Accuracy is not benefit
These studies measure whether a device counts correctly. They do not show that wearing one improves health. For that question see Fitness Bands and Activity Trackers.How to judge a step claim
- Ask which model and software version was tested.
- Ask who walked: age, health, speed, and what was counted as the reference.
- Look for the error size, not just "accurate".
The checklist in How to Read a Wearable Validation Study explains these terms.
Frequently asked questions
Can a phone count steps as well as a watch?
Phones use the same kind of sensor, but they are carried in different places and positions, which makes counting harder. [1] Results differ by model.
Are calorie estimates from a tracker reliable?
No. A systematic review found no brand accurate for energy expenditure. [2] Use them only as rough trends, if at all.
Do trackers over- or under-count steps?
Recent Fitbit models underestimated steps by about 3 per minute on average in a meta-analysis. [3] Other brands and models differ.
Do slow walking or walking aids matter?
Yes. Commercial wearables counted accurately only under specific conditions in a small study with walking aids. [5] Ask a clinician before relying on a device after surgery.
Does a high step count prove I am healthy?
No. Accuracy of counting is a separate question from health benefit, and a count is not a diagnosis.
More details: evidence, limits and cautions
The details behind this page: cautions, limits of the evidence and where sources disagree. Open any note to read it.
How to read the evidence labels
Labels used on this page: established (many reviews agree), mixed (results differ by device or condition), insufficient evidence (not enough studies on the situation).
Safety and common sense
References
- Song Z et al. Carrying Position-Independent Ensemble Machine Learning Step-Counting Algorithm for Smartphones. Sensors 2022. PubMed abstract. Read 2026-10-22.
- Fuller D et al. Reliability and Validity of Commercially Available Wearable Devices for Measuring Steps, Energy Expenditure, and Heart Rate: Systematic Review. JMIR Mhealth Uhealth 2020. PubMed abstract. Read 2026-10-22.
- Chevance G et al. Accuracy and Precision of Energy Expenditure, Heart Rate, and Steps Measured by Combined-Sensing Fitbits Against Reference Measures. JMIR Mhealth Uhealth 2022. PubMed abstract. Read 2026-10-22.
- Choe JP, Kang M. Apple watch accuracy in monitoring health metrics: a systematic review and meta-analysis. Physiol Meas 2025. PubMed abstract. Read 2026-10-22.
- Tsamassiotis S et al. Accelerometers can correctly count orthopaedic patients' early post-operative steps while using walking aids. J Exp Orthop 2025. PubMed abstract. Read 2026-10-22.
- Musa SB et al. Wearable Device Validity in Measuring Steps, Energy Expenditure, and Heart Rate Across Age, Gender, and Body Mass Index. J Phys Act Health 2023. PubMed abstract. Read 2026-10-22.
- Ramli AA et al. Gait Event Detection and Travel Distance Using Waist-Worn Accelerometers across a Range of Speeds. Sensors 2024. PubMed abstract. Read 2026-10-22.