Reaction time test
Runs entirely in your browser · no scores uploaded
Wait for the panel to turn green and click as fast as you can. Most people land between 200 and 300 milliseconds; the score shown is the median of your attempts, so a single lapse does not wreck it.
How to test your reaction time
What is actually being measured is a chain, not just you: light leaves the screen, your eye and brain process it, your finger moves, the mouse reports, and the browser timestamps it. A 60 Hz monitor adds up to 17 milliseconds of that on its own before anything else, and a wireless mouse can add several more. The same person therefore scores measurably faster on a 144 Hz screen with a wired mouse, because the hardware is genuinely part of the number. Treat the figure as a comparison against yourself on the same setup rather than against anyone else.
One part of that chain sits inside this page and is worth being clear about. The clock starts the instant the panel is told to turn green, not the instant it actually looks green to you. Between the two sits a browser render and a screen refresh, so every attempt carries one frame of the display in it, and the number you see is a little larger than your nervous system alone. Nothing here can remove that, because a web page has no way to know when a photon left the panel. It cancels out entirely when you are comparing two of your own sessions on one machine, and it does not when you are comparing yourself against somebody on different hardware.
The timing itself is not the weak link. Every measurement uses performance.now() rather than the wall clock, which is monotonic and cannot step sideways mid-test, and its resolution is 100 microseconds in an ordinary page. Human variation is four orders of magnitude larger than that. The median across attempts is used for the same reason: one lapse of attention produces a 700 millisecond outlier that would drag a five-attempt mean up by eighty milliseconds or more, and the median steps over it without being asked.
What people use it for
- Getting a baseline to compare yourself against on another day
- Seeing what a night of bad sleep is worth in milliseconds
- Checking whether a new monitor or a wired mouse changed anything measurable
- Warming up before something that rewards a fast response
- Settling an argument about who is quicker, on the same machine
- Watching what caffeine, or a long session, does to your own numbers
Questions
Around 270 milliseconds is the typical median for visual reaction. Under 215 is fast, and under 200 usually means anticipating rather than reacting.
They are fixed bands on the median: great under 215 milliseconds, good under 275, average under 350, and slow beyond that. A monitor’s own latency is inside those numbers, so the fastest band starts below anything nerve conduction could manage on its own.
Attention fluctuates far more than nerve conduction does, which is exactly the reason a median over several attempts means more than any single go.
Yes, noticeably. A 60 Hz screen can add up to 17 milliseconds before your eye sees anything, and panel input lag adds more on top.
It does. The clock starts when the page turns the panel green, and the frame carrying that change still has to be drawn and displayed. Compare yourself against yourself on one machine and the difference cancels out.
You clicked before green, or within 100 milliseconds of it. Nothing under 100 milliseconds is a reaction, so it was recorded as a guess rather than an attempt.
No. It is discarded rather than logged as a slow attempt, so guessing costs you nothing except the attempt.
Somewhere between 1.2 and 4.8 seconds, drawn fresh each time so it cannot be anticipated. A long wait is not the test misbehaving.
No. A click during the wait ends the attempt as a false start, and one inside the first 100 milliseconds after green does the same.
At least five. The median stops moving much after that, and ten gives a figure you can compare against another day with some confidence.
Because a single distraction produces an outlier several times the size of a normal attempt. The mean chases it; the median ignores it.
Far more accurate than the thing being timed. It uses performance.now(), which is monotonic and resolved to 100 microseconds in an ordinary page, against human variation measured in tens of milliseconds.
A few milliseconds, consistently. A wired mouse and a high polling rate both shave a little, so serious comparisons need identical hardware at both ends.
It can. Anything competing for the main thread delays the frame that turns the panel green. Close the heavy tabs if you are chasing a personal best.
No. This measures how fast you respond to a signal; a click speed test measures how many clicks you can make per second.
No. They live in the page while it is open and are gone when you close it. Nothing is uploaded and there is no leaderboard.