How to play
Click the pad once to arm an attempt. The pad turns amber and holds for a random interval between 1.2 and 3.5 seconds, then flips to green. Click again the moment you see green, and the gap between those two events is your time in milliseconds. Do that five times and the session closes with a summary.
The pad is strict about what counts, which is the difference between a test and a toy. Click while it is still amber and the attempt is voided, a short lockout stops you from simply spamming the button until green arrives, and that attempt is repeated rather than recorded. Any time under 80 milliseconds is also discarded and repeated, because a click that fast was already on its way down before the colour changed. Only five clean attempts make it into your numbers.
Four figures are reported while you go, and each answers a different question. Last is the attempt you just made. Best is your fastest of the five. Average is the mean of all five and is the one distraction ruins. Median is the middle value, which ignores a single disaster entirely. Your fastest single click and your best five-attempt average are stored on this device and shown under the pad the next time you visit.
Controls
- Mouse or trackpad: click the pad to arm, click again on green.
- Keyboard: Space or Enter do exactly what a click does, and held keys that auto-repeat are ignored so a stuck key cannot fake a fast time.
- Touch: tap the pad. It is deliberately large so that aiming is never part of what you are measuring.
- Try again: clears the five attempts and starts a new session. Your stored records are kept.
How to get a number that means something
The main thing separating a good session from a bad one is not reflexes, it is what you do with the waiting period. Two habits ruin more attempts than anything else.
The first is predicting instead of reacting. After a couple of attempts, your brain starts estimating when green is due and firing on schedule. It occasionally produces a spectacular number and much more often produces a false start, and the run of amber-click voids is the tell. The counter is boring but effective: keep your eyes on the pad and consciously refuse to guess. The 1.2 to 3.5 second window is wide precisely so that counting cannot pay off.
The second is pressure on the button. Resting your finger heavily on the mouse slows the click itself, because you have to lift before you can press. A light finger already touching the button, with the movement coming from the fingertip rather than the whole hand, is worth a measurable amount on its own.
Beyond that, the environment does more work than technique. Keep the tab in focus and close anything doing heavy work in the background. Full screen and a still page help, since anything animating is competing for the same frames. And run all five attempts in one sitting: a session where you paused halfway is not comparable with one where you did not.
Finally, read the median first. If your average is 245 and your median is 240, that is a real, repeatable figure. If your average is 280 and your median is 235, you had one bad attempt and the median is the number to believe.
Bands we measured on our own machines, which makes them estimates and not any kind of assessment: most adults using a mouse on an ordinary monitor land somewhere in the 200 to 280 millisecond range for the average of five. Below about 170 the test itself assumes you were anticipating rather than reacting. Consistency matters more than peak: a spread of under 40 milliseconds between your fastest and slowest attempt is a genuinely tidy session, and in ours a wired mouse ran roughly 10 to 20 milliseconds ahead of a trackpad on the same machine.
What a browser can and cannot measure
Timing a response to a signal is about as old as experimental psychology itself. The Dutch physiologist Franciscus Donders was timing responses to simple stimuli in the 1860s, and his subtractive method — timing a simple response, then a harder one, and treating the difference as the cost of the extra decision — is where the field of mental chronometry starts. Later work by William Hick in 1952 and Ray Hyman in 1953 established that choice reaction time grows with the logarithm of the number of options, which is why a test with a single pad, like this one, produces faster numbers than any test where you must decide which thing to press.
What your browser reports is not purely your nervous system, and it is worth being honest about the difference. The clock starts when the page paints the green pad and stops when the click event arrives. In between sit your display's refresh interval — on a 60 Hz screen a new frame is only shown every 16.7 milliseconds, so up to that much delay is added before you can possibly see the change — plus the polling rate of your mouse and the browser's own event handling. Those costs are consistent enough that comparing yourself against yourself on the same hardware is meaningful, and they are large enough that comparing your number to somebody else's on different hardware is not.
This test is our own code, written for this site. Timing uses the high-resolution clock rather than the wall clock, the random delay comes from your browser's cryptographic random source, and nothing you do here leaves your device. It is entertainment, not a diagnostic instrument, and no result on this page says anything about your health.
Questions players ask
Why was my fastest click thrown out?
Anything under 80 milliseconds is discarded and the attempt repeated. A click that early was in flight before the pad changed colour, so it measures a guess.
Why show a median as well as an average?
One distracted attempt can drag a five-attempt average up by tens of milliseconds while the median barely moves. When the two disagree, trust the median.
How long is the wait?
Between 1.2 and 3.5 seconds, redrawn randomly for every attempt so that counting in your head cannot help you.
Is this a medical test?
No. It is a browser measuring the gap between painting a colour and receiving a click, including your screen and input device. Compare it against your own past sessions and nothing else.