Does an under-desk bike or elliptical count as steps?

We went to Apple’s own support documentation on 2026-09-17 rather than describing this from memory, because the exact wording turns out to be the whole answer. Four statements, quoted:

• “The Activity app relies on arm motion and an accelerometer to track movement, but the Workout app can use the accelerometer, the heart rate sensor, and GPS.”
• “To make sure that you earn Exercise credit during walks, allow the arm with your Apple Watch to swing naturally.”
• “If you need both hands while walking, for example to push a stroller, you can still earn Exercise credit with the Workout app.”
• “Every full minute of movement that equals or exceeds the intensity of a brisk walk counts toward your daily Exercise and Move goals… Any activity below this level counts only toward your daily Move goal.”

Read those four together and the seated-machine problem is fully described by Apple before anyone mentions a pedal exerciser. Passive tracking reads your arm.Pedaling at a desk holds both arms still on a keyboard, which is a more extreme version of the stroller case Apple names. And even where motion is detected, the intensity threshold is “equals or exceeds… a brisk walk” — and pedaling at a pace you can type through is, by definition, not that.

The way out is in the first quotation. The Workout app can use the heart rate sensor, and your heart rate does not care where your arms are.

Fitbit’s help documentation, read on the same date, arrives at the same place from the opposite direction — it is mostly written about the tracker over-counting, and in doing so it states the mechanism exactly:

• “If you wear a wrist-based Fitbit device and move your body while keeping your arms still (or vice versa), you may see a slightly different step count.”
• “When working at a desk, cooking, or performing other activities with arm movements, a device on your wrist may add some steps.”
• “When you ride a bike on bumpy trails, your device may add some steps.”
• “Fitbit devices have a finely-tuned algorithm for step counting.”

The second and third quotations are the ones worth sitting with, because they say something slightly uncomfortable: a wrist tracker can add steps for typing and for riding over bumps. It is reading vibration and motion patterns at the wrist and inferring walking. Seated pedaling produces neither the arm swing of walking nor the jolts of a bumpy trail, so it produces very little of anything.

It is worth being clear that nothing here is broken, because the usual reaction is to assume the watch is wrong or the machine is faulty, and neither is true.

A wrist-worn tracker has one sensor for motion and it is on your wrist. It infers walking from a pattern of arm swing that has been tuned on people actually walking. Any activity that moves your legs while keeping your arms still is outside the pattern it was designed to detect: seated pedaling, but also cycling with your hands on the bars, pushing a shopping cart, and pushing a stroller — which is why Apple names that last one explicitly and tells you what to do about it.

What follows is a practical warning about a very common purchase mistake. If you buy a seated machine in order to close a ring, you will be disappointed within a week and the machine will stop being used. If you buy it to be less motionless during hours you were going to spend sitting anyway, it does that job, and we set out honestly what that is worth.

“Does a mini stepper count as steps” is a separate question with a better answer, and it is worth separating because the two machines get searched together.

On a stepper you are standing, and every step raises and lowers your entire body. That produces real vertical motion at the wrist whether or not you are consciously swinging your arms, so a tracker usually registers something. What it will not give you is a count that means what an outdoor step means. Apple notes that the watch calibrates the accelerometer by learning your stride length when you walk or run outside, and a stepper has no forward travel to calibrate against— you are going up and down on the spot.

And on the machine’s own side of the question there is nothing at all: no mini stepper we track publishes a step height, which means the work done per step cannot be computed even in principle. We took that arithmetic as far as it goes and then stopped, which is further than the listings do.

If your watch will not count it, the obvious next hope is the monitor on the machine. Here is what that monitor actually has to work with.

It knows revolutions and it knows elapsed time. It does not know what you weigh, how long your legs are, or which resistance setting you are on — not one machine we track takes any of those as an input. So a calorie figure on one of these displays is revolutions multiplied by a stored assumption about a person who is not you.

Syncing is also less automatic than most buyers expect. Cubii’s own model comparison, read on 2026-09-17, lists the app compatibility of the Move and the Go as “Manual Data Entry” — you read the numbers off the monitor and type them in yourself — and only the Total Body, the top of the range, as “Bluetooth Seamless Syncing”. Display follows the same ladder: “Basic LCD Monitor” on the Move against “Backlit LCD Monitor” above it.

Four checks, in order. Only the first two are usually stated in a listing, and the fourth is the one that actually solves the problem.

1. Does the machine sync, or do you type?Look for the words Bluetooth or app connectivity in the specification rather than in the marketing copy, and treat “app compatible” as ambiguous until you find out which of the two it means. Cubii’s own comparison page shows how far apart those can be inside a single brand’s range.

2. Does it publish a resistance magnitude, or just a count? If nothing can measure your output, the next best thing is knowing the input. A level count tells you how many positions the knob has. A stated resistance figure tells you what the hardest one is. Exactly one machine in this category publishes the second kind of number.

3. Can you put your watch into a workout instead? This costs nothing and is the single highest-value change on this page. Apple states in its own words that the Workout app can use the heart rate sensor rather than arm motion. Starting an indoor-cycle or other workout before you pedal moves you off the sensor that cannot see you and onto the one that can.

4. Measure the input the machine is missing. A chest strap reads your heart rate directly from your chest, which is the one signal that is genuinely about your body rather than about your wrist or about a stored assumption. If the reason you are on this page is that you want to know whether any of this is doing anything, this is the honest way to find out.

Against those checks, three machines are worth naming — two that publish app connectivity, and one that publishes the input figure instead.

Most resistance steps, with app tracking: the YOSUDA Under Desk Bike. Sixteen magnetic resistance levels, double the DeskCycle and the Cubii JR1, plus Bluetooth app tracking. The honest caveat is large and it is not about tracking: YOSUDA publishes no pedal height, which is the single figure that decides whether an under-desk machine fits under your desk, and no user weight limit either.

The cheapest way to find out whether you will use one: the Sunny mini pedal exerciser. Bluetooth app connectivity at the bottom of the price range, sold as a low-profile design for under-desk use, and it works for arms on a desktop as well as legs on the floor — which, given everything above about arm motion, is a more interesting feature on this page than on any other. Sunny publishes no pedal height, no dimensions and no user weight limit for it.

No app, and the only published resistance magnitude: the DeskCycle. It has no Bluetooth and we are recommending it anyway, because it is the only machine in this category that publishes eight magnetic settings up to 39 lb of resistance — a magnitude rather than a count. If nothing is going to measure your output reliably, knowing exactly what you are pushing against is the substitute that does not depend on a sensor.

Worth saying plainly, because this page could easily leave you chasing a metric that was never a good fit.

A step count became the number everyone tracks because it is easy to collect from a wrist, not because it measures work. Public health guidance is not written in steps. MedlinePlus, read on 2026-09-17, states it in minutes: “Get at least 150 to 300 minutes of moderate-intensity or 75 to 150 minutes of vigorous-intensity aerobic physical activity each week”, and defines the threshold by a test you can run without any device at all — “moderate intensity means that while you are doing that activity you should be able to say a few words in a row but not sing”.

Apply that talk test to a seated machine at a pace you can work through and it usually comes out below moderate. That is a more useful and more honest read of what you are getting than any step count would be, and it is available to you for free, right now, on the machine you already own. We ran the same test on walking pads and it reaches a similarly unflattering conclusion about desk pace.

We have not tested a watch against any of these machines. No wearables were compared, no step counts were logged, and nothing on this page is a measurement we made. What we did was read Apple’s and Fitbit’s own published documentation on 2026-09-17, quote it exactly, and reason from what it says about how the sensors work. Our full method.

We have not given you a conversion from revolutions to steps, and you should be suspicious of anyone who does. It would require a stride length, a resistance figure and your body weight, and the machines publish none of the three.

This is general information about published specifications and published vendor documentation. It is not medical advice.