Reaction Time Test

This is the simplest kind of reaction test. One thing happens, you make one response, and there is nothing to decide. The clock starts when the screen lights up. It stops when your mouse button closes.

Four things sit inside that gap. Your eye catches the light. The signal travels to your brain. Your brain tells your hand to move. Your finger moves. All four are in your score. Thinking is not, because there is nothing here to think about.

The test comes from Franciscus Donders, a Dutch scientist working in 1868. He timed simple responses like this one. Then he timed harder ones and subtracted the first from the second. What was left over was the time the extra thinking took. That trick is why we publish the four-choice version alongside this test. On its own, your score here is a floor. It is the fastest you can turn a signal into a movement.

Your five trials will not look alike. Four will land close together, and one will land much slower because your attention slipped for a moment. That slow trial is the problem.

Average the five and one slow trial drags the result up by around 80 ms. The median barely moves. The median is the middle value: line your trials up from fastest to slowest and take the one in the middle. It ignores a stray result almost completely.

That is the whole reason this test reports a median. It is also why five trials is the minimum and twenty is offered in settings. Five trials give a median worth keeping. Twenty give one steady enough to compare across days.

First, we start the clock when the light actually appears on your screen. That sounds obvious. It is not what every site does. Code can decide to show the light one frame before the screen draws it, and a frame lasts 16.7 ms on a 60 Hz display. Start the clock at the decision instead of the paint and every score comes out 16.7 ms too fast. We start it at the paint.

Second, we throw away any response faster than 100 ms and tell you why. Nobody sees a surprise light and reacts in under 100 ms. A press that fast had already started before the light appeared. You guessed, and the guess landed well. These two rules cost an honest tester nothing. They cost a guesser a lot.

A browser test measures you plus your equipment. Your screen takes time to show the light. Your mouse takes time to report the click. Your operating system takes time to pass it on. Together that adds roughly 30 to 80 ms, depending on your hardware.

So your real reaction time is faster than the number on this page. We could quietly subtract a flattering amount and hide that. We do not. The methodology page sets out what we correct and what we leave alone.

This is why comparisons only work between numbers measured the same way. Two questions are worth asking. Are you faster than you were last week on the same machine? And where do you sit among everyone else testing in a browser? For the age picture, average reaction time by age has the curve. What counts as a good reaction time explains why the popular 250 ms figure is folklore rather than finding.

Riding the click. Holding the button half pressed while you wait does not make you faster. It makes you a false starter, because the filter catches the release pattern it produces.

Chasing one lucky trial. We show your fastest single trial because people want to see it. It is the least reliable number on the page. The median is your score.

Switching input devices between runs. A trackpad and a mouse differ by more than any improvement you are trying to spot. Pick one and stay with it.