what the iPhone counts as a step
One rule answers every version of this question: the phone counts a step when it feels a peak in the acceleration of the body carrying it. Not your legs — the phone. Everything below follows from that, including the cases where the answer is no.
What has been measured
Short list, because it is a short list. These are trials, not reasoning.
| situation | counts? | what was found |
|---|---|---|
| Treadmill, phone on you | yes | within 5% of hand-counted steps at 5, 7.5 and 10 km/h |
| Ordinary days, phone in a pocket | yes | 791 steps a day missing against a reference pedometer |
| Ordinary days, phone in a bag | partly | 2,104 steps a day missing — the bag costs about 1,300 |
What follows from how it works
Nobody has published a trial for the cases below. Rather than guess at percentages, this table says what the mechanism implies and leaves it at that.
| situation | counts? | why |
|---|---|---|
| Phone on the treadmill console | no | you move, the phone doesn't |
| Walking in place | usually | your body still rises and falls |
| Elliptical | usually not | a smooth glide, no heel strike |
| Cycling | no | no heel strike, and Apple doesn't track it |
| Phone switched off by you | assume not | no documentation either way |
| Battery flat, phone shut itself down | sometimes | reserve charge — reported, not documented |
| Pushing a wheelchair | not as steps | a different movement, counted differently |
| Rough road in a car | may add extra | vibration looks like a walking rhythm |
The treadmill, and the mistake everyone makes on it
A treadmill is not the problem. Carried on your body, the phone matched hand-counted steps to within 5% at three different speeds. The problem is the console: it is exactly where a phone ends up, and a phone lying still counts nothing while you walk five kilometres.
Holding the handrails costs you as well. The motion that reaches your pocket gets smaller, and smaller motion means fewer peaks that clear the threshold.
Cycling
No — and for two reasons that agree. Apple's Health app records steps and walking or running distance by itself; cycling is not one of the things an iPhone tracks on its own, and it needs a workout app or an Apple Watch. It also could not work: pedalling has no heel strike, so there is no peak for the detector to see.
When the phone is off, or the battery is dead
These are two different states, and the second one is stranger than it looks.
Switched off by you: assume nothing is recorded. Neither Apple nor anyone else documents what the motion chip does after a deliberate shutdown, and this page is not going to guess.
Battery flat: an iPhone does not run to zero — it shuts itself down while a little charge is left. There are documented accounts of the motion coprocessor continuing to log steps for days after that, with the count appearing once the phone was charged. Apple does not describe this as a feature, the best-known account is from an iPhone 5s in 2014, and nothing says today's hardware behaves the same way. Do not rely on it — but it explains the occasional surprise of steps turning up for hours you had written off.
This section was wrong when the page was first published: it said flatly that nothing is recorded in either case. That is true of a chip with no power at all, and it is not true of a phone that shut itself down with charge to spare. Corrected the same day.
Walking in place, and the elliptical
Walking in place usually counts: your body still rises and falls the way it does on a pavement, and that is the whole signal. An elliptical usually does not: the motion is a glide, your feet never strike anything, and the sharp change the detector looks for never arrives.
Both of those are reasoning, not measurement. No published trial has tested an iPhone on an elliptical, and this page would rather say so than invent a number.
Wheelchairs
Pushes are not steps, and Apple treats them as their own measurement rather than converting them into steps: on Apple Watch, wheelchair users get a count of pushes, with a choice of walking pace or running pace. An iPhone alone does not track pushes.
Extra steps in the car
A rough road shakes the phone at about the rhythm of walking, and a step counter is built to recognise exactly that kind of repeated jolt. It is not a fault so much as the right detector meeting the wrong input.
There is no published figure for how many steps this adds on an iPhone, so there is none here either.
What this means for the number on your screen
Angel Step shows the count Apple Health already holds, so it inherits exactly these rules — no more and no less. If the number looks lower than your day felt, the reason is usually in one of the tables above; if it disagrees with another device, why your step counts don't match explains which one Health keeps. And how accurate the count is answers the other half of the question: not whether a movement registered, but how close the total lands.
Sources
- Apple, Core Motion — the motion coprocessor records movement in the background and keeps seven days of history. Apple Developer
- Reported behaviour after a flat-battery shutdown — an iPhone 5s logging for four days on reserve charge. ⚠️ A single old account, not Apple documentation. iDownloadBlog, 2014
- Apple — wheelchair users get a push count with a choice of pace. Apple Support
- Duncan et al. (2018), Journal of Sports Sciences — treadmill and free-living validity of the iPhone step count. PubMed
- Höchsmann et al. (2019), JMIR mHealth and uHealth — pocket against bag over seven days. JMIR