Reaction Time Test

Press start and the lever goes down. After a random wait of 1.5 to 6 seconds the toast pops: the dark stage turns bright yellow, the toast jumps and POP! fills the screen. Press as soon as you see it, with a mouse button, a finger or the spacebar, which also starts the test. Do that five times to get your average and your best.

A press before the pop is a false start, and that slice just runs again. So does a press under 100 ms, because that is too quick to be a response to what you saw. The test sleep researchers use counts responses under 100 ms as false starts, a large 2015 study threw out everything under 110 ms, and in sprinting a start within 0.100 seconds of the gun is a false start too.

The wait is random in a particular way: however long it has gone on, the pop is about as likely in the next moment as it was at the start. People respond faster when they can tell a signal is due, so a predictable wait would reward timing rather than reacting.

There is no single correct figure, because every number depends on the equipment. In that 2015 study, 1,469 adults aged 18 to 65 averaged 231 ms clicking a mouse at a target on a 60 Hz screen, and about 18 ms of that was the equipment's own delay. The same paper lists other computer-based studies with averages from 233 ms to nearly 400 ms. Human Benchmark, another browser test, publishes a median of 273 ms from over 81 million clicks.

So on an ordinary computer, about 200 to 300 ms is normal for most adults. The study's averages by age, on its lab equipment, show how gentle the slope is: read them as a shape, not a target.

Age. People slow down with age, but gently. The 2015 study found about 0.55 ms a year between 18 and 65, and a long-running Baltimore study about 0.5 ms a year from around age 20, in a test using sound. A survey of 7,130 UK adults, though, found little slowing until around 50. The studies agree the change is small and disagree on when it starts.

Tiredness. Sleep loss has one of the clearest effects. The vigilance test used in sleep research, a longer version of this kind of test, shows slower responses after 33 hours awake and after five nights of four hours' sleep. In another study, two weeks of four or six hours a night built up deficits while people's own ratings of their sleepiness rose only slightly after the first days.

Your device and screen. Every test on your own device includes that device's delay. In a lab test with a light sensor and a robot finger, browsers overestimated reaction time by 58 to 70 ms on two phones and by 62 to 133 ms on two laptops, and the figure changed with the browser. Human Benchmark puts a typical computer at about 30 ms and says taps tend to run a bit slower than clicks. So phones often read slower in everyday use, though that lab's phones were no worse than its laptops. The sources disagree on how much, and agree that devices differ, so scores compare best on the same device and browser.

The honest answer first: simple reaction time does not move much with practice. In a two-year study, baseball practice improved players' times when they had to decide whether to respond, but not their simple reaction time, and the sleep researchers' test does not improve with repeated testing. The evidence is not all one way: a small three-week training study did find faster times. One review credits action video games with faster responses, a later experiment found no faster processing in simple tasks, and other researchers question how the earlier studies were designed.

What you can do is give yourself a fair test. Sleep first. Look straight at the stage, rest a finger on the button before you start, and do not guess. Five trials is a small sample, so take a few tests and compare averages, on the same device.

If you would rather play than measure, Butterflinch is our daily reaction time game, with tricks to make you flinch. Target Tap is a desktop aim trainer that also reports your average reaction time, Pop Pulse is an aim and timing game of fireworks that burn down, and Hold Fast is a timing game for several fingers on a touchscreen. There are more on our reaction games page.

The start time is the screen frame that shows the signal, and the end time is the moment your browser received your press. Both are on the same clock, so results are not rounded to screen frames the way a game running 60 steps a second would be, and a busy page does not lengthen a result, because the press carries its own time.

We test this with scripted presses at known moments, on the page's real code, in Chrome, Firefox and Safari's engine. Typical results came back about 2 ms after the true time in Chrome and Safari's engine and 3 to 6 ms after it in Firefox, and the worst was 19 ms, the scripting's own delay included. With the page deliberately frozen for 300 ms they stayed within 8 ms. The browsers' clocks step 0.1 ms in Chrome and 1 ms in the other two. Chrome also reports when a frame was presented: about 12 ms after our start time on our test machine, so a result there reads about that much high.

No web page can see your screen, mouse or touchscreen. The moment the signal really appears is known only to the nearest frame or two, one frame being about 17 ms at 60 Hz and 8 ms at 120 Hz, and the device delays above come on top. Treat a 20 ms difference between two runs as noise.