Guides · Watch vs treadmill
Why your watch and your treadmill never agree (and which one is right)
You finish a run, and the treadmill says 5.6 miles. Your watch says 5.4. Which one do you trust? Maybe neither. Both can float above what the belt really did.
Quick answers
Why do my watch and treadmill show different numbers? They're built to measure different things in different conditions, and neither one is reading the belt directly. Your watch has no GPS indoors, so on a treadmill it's estimating from your arm swing and stride, not tracking your position. Your treadmill knows how fast its motor is trying to spin the belt, not how fast the belt actually moves under your feet.
When my watch and treadmill agree, does that mean they're both accurate? Not necessarily. Indoors your watch is estimating, and some watches pull in the treadmill's own speed signal if paired, so agreement can mean the watch is echoing the treadmill's error rather than confirming it.
Which one should I trust, my watch or my treadmill? Neither, by default. Both are estimates. The only way to know your real pace is to measure the belt itself.
Does my watch use GPS on a treadmill? No. GPS doesn't work indoors. Your watch switches to its accelerometer and gyroscope to estimate distance from your movement, which is a rougher method than satellite tracking.
The number you're actually staring at
You have never actually known your treadmill pace. Nobody has. The number on the display is what the motor was aiming for, not what the belt is really doing under your feet. When your watch matches that number, it feels like confirmation. It usually isn't.
| Device | What it actually reads | Is it watching the belt? |
|---|---|---|
| Your watch, indoors | Arm swing and stride rhythm, sometimes the treadmill's own reported speed | No |
| Your watch, outdoors | Satellite position, 3% to 6% mean distance error in published testing | No belt involved |
| The treadmill display | The speed the motor is commanded to hold | No |
| A contact tachometer | The belt's true surface speed | Yes |
| Treadwell on an iPhone Pro | The belt's true surface speed, by camera and LiDAR | Yes |
What your watch is really doing indoors
Outside, your watch tracks your position with GPS and calculates distance from that. Indoors, GPS signals can't reach your wrist, so the watch switches to a different method entirely: it reads your arm swing and stride rhythm through its accelerometer and estimates distance from that movement. Some watches also pull in the treadmill's own reported speed if you pair them, so the watch ends up repeating the treadmill's number instead of checking it.
Even outdoors, where GPS actually works, watches aren't as precise as people assume. A peer-reviewed study of eight popular sport watches (Apple, Coros, Garmin, Polar, Suunto) found mean distance errors of 3% to 6% against a known course. That's outdoors, with real satellites. Indoors on a treadmill, there is no GPS at all, so the watch is working with less information, not more.
Two different jobs, one watch
How your watch turns a run into a distance number, in three steps. Outside it measures. On a treadmill it guesses.
Outdoors: measured in 3 steps
↓
↓
Still 3% to 6% off in published testing.
Indoors: guessed in 3 steps
↓
↓
No satellites, no belt reading, no way to check itself.
Neither path touches the belt, which is the only surface your feet actually move against.
What your treadmill is really doing
The treadmill's display isn't reading the belt's speed either. It's a command readout. When you press 6.0, the display shows 6.0 because that's what the motor was told to do, not because anything measured the belt doing it. Most machines have no sensor watching the belt at all, so the number on the screen never changes even as the machine drifts away from it.
Machines drift, too. A 2021 study in Scientific Reports tested belt speeds against their set speeds and found even well-maintained machines running between 98% and 103% of target. Worn belts, loose belt tension, aging motors, and machines that have never been calibrated push the gap much wider, and the display has no way to know. That's how a machine ends up showing 5.6 miles for a 5.0-mile run while its screen looks perfectly confident.
There's no reliable published average for how far off machines run in the real world, because nobody audits gym equipment. What we've measured ourselves shows the range is wide.
Two runs where we caught it happening
We test treadmills against a handheld tachometer, a tool that reads the belt's true surface speed directly, not what the display reports.
| Machine | Display said | Really was | Gap |
|---|---|---|---|
| Gym treadmill | 5.6 mi | 5.0 mi | +12% |
| A different gym treadmill | 2.3 mi | 2.0 mi | +15% |
Same run, three different answers
A typical indoor run. The display overstates, the watch estimates a little lower, and neither reads the belt.
Illustrative run. Both numbers float above the real one, and neither device knows it. Scale starts at 4.0 mi to show the gap.
In our own testing, one machine's display claimed 5.6 miles for a run our belt measurement put at 5.0. The watch was no help catching it. Indoors a watch estimates from your stride, or repeats the treadmill's own signal if paired, so its number can land anywhere near the display's, a little under it, sometimes right on it. Close to the display never means close to the belt.
On a different machine, the treadmill displayed 2.3 miles for a run that was really 2.0, a 15% inflation the runner had no way to see from the display alone.
Not every machine is far off. In published testing, well-maintained treadmills ran within a couple percent of their set speed. The real problem: you can't tell which machines are off just by looking at the number.
So how do you actually find out?
Compare against a measured outdoor distance. Run a known distance outside by GPS, note the effort, then match that same effort on the treadmill and see what it claims. This catches big errors but not small ones, since matching effort by feel is a shaky ruler.
Use a contact tachometer. A handheld tachometer reads belt surface speed directly. This is what we use as ground truth. It's accurate, but it's a tool you have to buy and hold against a moving belt, which isn't something you do mid-run.
Measure the belt yourself. Measure the belt's length, mark a spot on its side, run the treadmill at a set speed, and count belt revolutions over a timed minute. Length times revolutions gives you true belt speed. It works, but you're timing belt laps with a stopwatch instead of running.
Use your phone's camera. Treadwell measures the belt's true speed with your iPhone's camera and LiDAR. You hold the phone over the running belt for a few seconds and it reads the belt itself, not the display and not your stride. In our tachometer tests, a reading on a machine it had never seen landed within half a percent of true. It needs an iPhone Pro, because those have the LiDAR sensor it measures with.
Settle it with the belt
Measure any treadmill's true speed in seconds, with just your iPhone.
What to do when they disagree
If your watch and treadmill are close, that only means they're not far apart. It doesn't prove either one is right. If they're way off from each other, at least one of them is wrong, and it could be both. The only way past the disagreement is to measure the belt directly instead of trusting either estimate.
Your watch is guessing. Your treadmill is reporting a command, not a measurement. Neither one is watching the belt.
Sources.
- Belt speed vs set speed: Willwacher et al., Scientific Reports 11:2244 (2021). Belt speeds ran 98% to 103% of target on well-maintained machines.
- Watch distance accuracy: JMIR mHealth & uHealth (2020). Eight positioning-enabled sport watches, 3.2% to 6.1% mean distance error outdoors (the 3% to 6% in the text, in round numbers).
- Treadmill and watch measurements described are our own tests. The 2.0-mile run and the never-before-seen machine reading are tachometer-verified.
- The bar chart is an illustrative example, not a logged test.