Reaction Time Test

Wait for the panel to turn green, then click or press Space as fast as you can. Five rounds give a median, average and best time in ms.

Every timing happens locally: random delays use your browser's crypto source and no result ever leaves the page.

—

You can press Space instead of clicking; the panel also works with touch.

How It Works

The test is a three-state loop. You click the panel (or press Space) to arm it; it turns gray and waits a random interval between 1.5 and 5 seconds, drawn from your browser's cryptographic random source so no pattern is learnable. When it flips to green, a timestamp is taken; your click takes a second one, and the difference in milliseconds is the round's result. Clicking during the gray wait is scored as too early and the round simply restarts — early clicks are never counted as fast ones. Five valid rounds complete a set, and the panel then shows your best, average and median together with all five times so you can see the spread, not just a headline number.

What reaction time measures
This is a "simple reaction time" task: one stimulus, one response, no decisions to make. It bundles the whole chain — retina to visual cortex, motor planning, nerve to muscle, key or button travel — and alert adults typically land around 250 ms to a visual stimulus (hearing is a little faster, ~170 ms, and pain fastest). Anything requiring a choice (which key to press) is "discrimination reaction time" and grows with the number of options, per Hick's law.
Why the median, not just the mean
Reaction distributions have a long right tail: attention lapses cost hundreds of milliseconds, and no equivalent penalty exists on the fast side, so the average is pulled upward by one bad round. The median of five takes the middle value, discarding the best and worst tails, which makes it a far steadier estimate of your typical speed — it is also the statistic labs report for small round counts.
Anticipatory clicks
Because the wait is uniformly random, the hazard rate is flat — clicking early is never a good bet — but the brain likes to guess anyway. A "reaction" under about 100 ms is physiologically almost impossible for a visual stimulus, so scores like that are anticipations: the click fires just before or during the change, driven by expectation. The gray wait state rejects clicks that beat the green; that is why "too early" costs you a retry rather than a flattering number.

Frequently Asked Questions

Can I actually improve my reaction time?

Partially. Simple reaction time has a large inherited component and it slows with age, but the task itself shows a practice effect of roughly 10–15% over the first few sessions — mostly from learning to wait attentively instead of drifting. Sleep, body temperature and alertness move the number day to day more reliably than any training routine does.

Does my screen or browser limit the measurement?

Yes, and this is a real caveat. On a 60 Hz display the green state can reach your eyes up to about 10–17 ms after the browser decided to show it, because painting happens at frame boundaries; LCD pixel response and the OS compositor add a few more milliseconds of lag that the timer cannot see. A value of 240 ms on this page is closer to 220 ms on a 240 Hz monitor, and to a lab rig it is closer still. The test is a fair comparison between your own runs, not a certified instrument.

Is a sub-200 ms score suspicious?

Usually, yes. Genuinely under-200 ms responses on a simple visual task are rare among untrained adults, and on a browser test they are more often anticipatory: you clicked on the color you expected, milliseconds before it appeared, and the timer credits you for a guess. This test rejects early clicks; it cannot see a click that lands 80 ms after the change driven by expectation rather than reaction. Treat 5-round medians, not single lucky rounds.

Why do you show the median as well as the average?

Because reaction times are lopsided: one 600 ms lapse of attention inflates your average while telling you nothing about typical performance, and one 190 ms fluke deflates it. The median — the middle of your five rounds — ignores both tails, so it is the sturdier summary of your real speed and the number comparisons should be based on.

Is anything uploaded to a server?

No. The random waiting interval is drawn from your browser's own crypto random source, and the timing is a plain performance.now() difference computed on this page. No result is transmitted, stored, or compared against other visitors.