Reaction Time Test

How to play

1. Click, tap, or press Space on the big panel to start. The box turns red.
2. Wait. After a random delay of between 1.5 and 5 seconds the box turns green.
3. React — click, tap or press Space the instant you see green. Your time in milliseconds appears immediately.
4. Click again for the next attempt. A round is five attempts.
5. After the fifth attempt you get your average, median, best and worst, a bar chart of the round, and an emoji rating. Press 🔁 Play again for another round.

Click before the green appears and you get a friendly ❌ Too soon! message. That attempt does not count and is not added to your times — you simply retry it, and the false-start counter ticks up so you can see whether you are reacting or guessing.

The whole test is keyboard accessible: Space and Enter do everything a click does, state changes are announced to screen readers through a live region, and if your system asks for reduced motion the colour change happens without any transition or flashing.

Scoring

Every timing uses performance.now(), a high-resolution monotonic clock that is not affected by the system time changing. The stopwatch starts the moment the browser is told to paint green and stops on your input event.

The emoji rating uses the trimmed average: ⚡ under 200 ms, 🚀 200–249 ms, 🏃 250–299 ms, 🚶 300–399 ms, and 🐢 at 400 ms and above.

Why three different averages? Because a single distracted attempt distorts a five-sample mean enormously. If you score 220, 240, 230, 250 and then 1800 because you looked away, the plain average is 548 ms — a number that describes nothing about your reflexes. The median is 240 and the trimmed average is 235, both of which actually reflect how you reacted. Watch the trimmed figure if you want a stable measure of progress, and the best figure if you are chasing a personal record.

What the numbers mean

Reaction time is not one thing. It is a chain: light leaves the screen, hits your retina, gets converted into nerve signals, travels to the visual cortex, is recognised as a change, triggers a motor command, travels down your arm, and finally your finger moves far enough to close a switch. Each link costs time, and the visual links are the expensive ones — which is why audio reaction times are typically 20 to 40 ms quicker than visual ones.

General figures put simple visual reaction time at roughly 200 to 300 milliseconds for healthy adults. Sprinters at the Olympic start line are held to a 100 ms rule precisely because reacting faster than that to a starting pistol is considered physically implausible; it is scored as a false start.

Your measured score also includes everything between the software and your finger. On the display side that means the browser’s paint timing and your monitor’s refresh interval — at 60 Hz a frame is 16.7 ms, so up to that much can be added just waiting for the next refresh. On the input side, mouse polling, trackpad smoothing and touchscreen sampling all add a few milliseconds each. None of that is your nervous system, and all of it is included in the number you see. It is another reason the only fair comparison is against your own previous rounds on the same setup.

Tips to improve

• Rest your finger on the button. Hovering costs you the distance. Light contact, no pressure, ready to press.
• Look at the middle of the panel, not the edges. Peripheral vision is quick at detecting motion but slower at detecting colour change over a large area.
• Do not try to predict the timing. The delay is randomised between 1.5 and 5 seconds specifically to punish anticipation. Guessing produces false starts and, worse, unstable scores.
• Warm up. Your first round of the day is usually your slowest. Play two rounds and judge yourself on the second.
• Use a mouse if you have one. Trackpads and touchscreens add measurable latency compared with a physical button.
• Check the obvious. Tiredness, alcohol and sitting slumped all cost tens of milliseconds. Being well-rested and sitting upright is worth more than any technique.
• Track the trimmed average, not the best. One lucky 180 ms tells you very little; five consistent attempts tell you a lot.

Privacy

The test runs entirely in your browser. There is no account, no upload and no server involved in the timing — performance.now() is read locally, the random delay is generated locally, and the result never leaves your device. Your best-ever time and your last ten attempts are kept in this browser’s localStorage so your records survive a refresh; clearing your site data, or pressing 🧹 Clear saved records, removes them permanently. The page also works offline once it has loaded.

Last reviewed: September 21, 2026