How the timing works
A round starts when you press. After a random wait of 1.5 to 4.5 seconds, a timer hands the change to the next screen refresh. The browser reports the time of that frame, and the page treats it as the moment the signal appeared. When you press, the page reads the time stamp the browser put on your input event and subtracts the frame time. The result is shown as a whole number of milliseconds.
A browser does not say where in the chain it stamps an input event, so this page does not claim that the stamp removes input lag. It does not subtract any delay for your screen, mouse or browser.
The score is the median of five rounds. No round is dropped, so one slow round cannot hide, and one lucky round cannot decide the result. The page also shows the best round, the worst round, the mean and the spread (worst minus best). A spread over 100 ms adds a note that your rounds varied a lot; that threshold is ours.
The rules for a round
- Too early. A press before the signal, or an event stamped before the signal frame, does not count. The round restarts after your next press, and the false start is counted.
- Under 100 ms. A press less than 100 ms after the signal is treated as a guess, not a reaction, and the round repeats. This cut-off is ours, and it is a choice.
- Missed. No press within 3 seconds after the signal is a miss, and the round repeats. Three misses in one run end the run.
- Hidden tab. If the tab is hidden while a round is waiting, the round is canceled and starts again when you return.
Where the number can be off
The number is you plus your screen, your input device and your browser. It is useful for comparing yourself on the same setup. The sizes below are not measured on your setup.
| Source | Size | Notes |
|---|---|---|
| Display wait | Up to one frame: 16.7 ms at 60 Hz, 6.9 ms at 144 Hz, 4.2 ms at 240 Hz | Exact arithmetic (1000 divided by the refresh rate). The signal can only appear on a screen refresh. The test shows the refresh rate it measures on your screen. |
| Input device | Varies by device | Mice, keyboards, wireless links and Bluetooth can add delay of their own. We do not know its size on your setup and the page does not subtract it. Display, input and browser together are part of every number, which is why it suits comparing yourself with yourself. |
| Browser | Timer rounding | Firefox with privacy.resistFingerprinting turned on rounds timers to at least 16.667 ms. |
| Touch screen | Varies by device | A touch screen has its own sensing delay, so the same person can get a different number by touch than with a mouse. We have no figure for how much, so we give none. |
Because of these sources, we do not say the result is accurate to a given number of milliseconds, and the page never calls it a clinical reflex. A "reflex test" at a clinic is a different exam.
Our bands
These bands are ours, chosen to make a number readable. They are not a percentile and not a published norm. They are our own choice of cut-offs, not data from other people.
| Median | Band |
|---|---|
| under 150 ms | Very fast |
| 150–199 ms | Fast |
| 200–249 ms | Moderate |
| 250–299 ms | Slower |
| 300–399 ms | Slow |
| 400 ms and up | Very slow |
This page quotes no average reaction time. Averages differ by source and by how they were measured; wherever this site shows one, it names the source.
Method versions
The current method is v1.0, reviewed on Nov 8, 2026. When a method changes in a way that can change results, the version number goes up and the date changes.
- v1.0, Nov 8, 2026: first release.