This is a simple visual reaction time benchmark, stripped down to the one thing it's actually measuring. The screen turns red — then, after an unpredictable delay, it flips to green. Tap the instant you see green. Do it five times in a row and you'll get your median reaction time in milliseconds, along with a rank showing where that sits among typical human performance, plus your fastest and slowest attempts shown side by side so you can see how consistent you were across the run.
Tap or click anywhere on the screen, or press Space, to start each attempt. Wait for green — the delay before the flip is randomised between roughly 1.5 and 4.5 seconds specifically so you can't learn the rhythm and start anticipating it after the first attempt or two. Tapping while the screen is still red is recorded as a false start, and that attempt doesn't count toward your five — you simply retake it. The test runs five attempts total, and the results screen only appears once all five have been completed cleanly.
Most reaction tests report your single fastest attempt, which mostly measures how lucky you got on one try rather than your actual reflex speed. This test instead reports the median of five attempts — the middle value once they're sorted — which is far more resistant to both a fluke fast tap and one distracted slow one dragging the number around. A single lucky 180ms tap buried among four ordinary attempts around 270ms won't be reported as your score, and neither will a single slow 400ms tap where your attention lapsed for a moment. The median is a much more honest description of your actual, repeatable reaction speed than any single best try could be. Your fastest and slowest attempts are still shown separately in the results so you can see the full spread, not just the summary number.
For a simple visual stimulus like this, most adults land somewhere between 250ms and 300ms. Under 220ms is genuinely fast, and under 190ms is exceptional and close to the physical limits of human visual-motor response. Above 350ms usually points to fatigue, distraction, or input lag rather than anything meaningful about your underlying reflexes, so a single slow session isn't worth reading much into. It's also worth remembering that a meaningful chunk of your measured time isn't really "you" at all: it's your screen's refresh interval, your browser's own event-handling overhead, and your mouse or touchscreen's own latency, all stacked on top of your actual reaction. Testing on a 60Hz display, for instance, can add up to 16 milliseconds before your finger even has a chance to move, simply because the green frame itself can only appear on one of that display's fixed refresh ticks.
The stopwatch starts on the exact frame the green is painted to the screen, not at the moment the game internally decides to switch colors — if timing started at that internal decision instead, you'd effectively be timed from a moment you couldn't yet see on screen, and every recorded score would read slower than your reflexes actually are. Timing itself uses the browser's high-resolution monotonic performance clock rather than the ordinary wall clock, so a background system clock adjustment or daylight-saving shift happening mid-test can't corrupt a result the way it could with a naive timestamp difference.
Your all-time best median is saved on your device, separate from the five-attempt result of your current session, so you always have a personal baseline to measure a new attempt against rather than just a single-session number that resets every time you reload the page. Reaction time is naturally noisy from one sitting to the next — factors like tiredness, caffeine, screen brightness, and even how relaxed your hand is on the mouse all shift the number by tens of milliseconds — so a single run below or above your usual range isn't necessarily meaningful on its own. Tracking your best over several sessions, taken at different times of day, gives a far more reliable picture of your real reaction speed than treating any one run as definitive.
Rest your finger lightly on the screen, or your finger on the mouse button, before the flip happens — the physical travel time from a resting hand to a click still counts against your time. Look steadily at the centre of the screen where the color change happens rather than letting your eyes wander or scan the edges, since noticing the flip a fraction of a second later than you could have is a common and avoidable source of slow attempts. And don't try to guess the timing: an anticipated tap that lands even slightly before the flip is a false start and gets discarded, while one that happens to land right as the color changes isn't really measuring your reflexes at all, just how well you predicted the randomised delay.