Three modes
- Visual is the classic: wait for green, hit. This is the one with the leaderboard, the percentile and the age comparison.
- Sound replaces the colour change with a beep. Nothing on the screen moves, so you can close your eyes. Most people are 30 to 50 ms faster to a sound than to a light: published reviews put simple reaction time at 140 to 160 ms for sound and 180 to 200 ms for light under lab conditions, because the ear gets its signal to the brain in about 10 ms and the eye needs 20 to 40.
- Go / no-go shows green or purple. Green, you hit. Purple, you keep still for a second. It measures how fast you are when you have to look before you move, which is closer to what a game or a road asks of you. Expect it to come out clearly slower than your plain visual score, often by 60 to 100 ms, and count the purple mistakes as part of the result.
Why the median and not the average
One distracted round at 600 ms wrecks an average. It barely moves a median. Five rounds is enough to throw away the outliers and still finish in under a minute, which is about how long anyone's patience lasts for this.
A false start (hitting before the signal) doesn't count as a round; you repeat it. The number of false starts is shown with the result because it says something too: lots of them means you were guessing the timing, not reacting.
What the milliseconds mean
| Median, visual | Level | Notes |
|---|---|---|
| under 180 | Elite | Rare on a normal monitor and mouse |
| 180 to 220 | Fast | Gamers with a high refresh screen |
| 220 to 260 | Average | Where most people land |
| 260 to 320 | Relaxed | Tired, distracted, or a slow setup |
| 320+ | Sleepy | Try again after a walk |
Anything under about 120 ms is not a reaction. The signal has to travel eye to brain to finger, and that alone takes longer. A score down there means you anticipated the green or a device is clicking for you.
"Faster than N%" under the result is an estimate. Until enough runs have been posted here it comes from a curve fitted to the figures Human Benchmark publishes for its own visitors, a median of 273 ms and a mean of 284. Once this site's reaction board passes 300 posted runs, the percentile switches to those and says so.
Average reaction time by age
The best numbers I know of come from a 2015 study by Woods and colleagues: 1,469 people aged 18 to 65, tested on the same hardware. Mean simple reaction time was 231 ms, or 213 ms once the measured delay of their screen and mouse (17.8 ms) was taken out. Age slowed it by 0.55 ms per year. That's all: about 22 ms between an 18-year-old and a 58-year-old.
Online tests run slower than a lab because every home screen, mouse and browser adds its own delay, which is a large part of why a 213 ms lab mean and a 273 ms website median can both be right. Put that slope on top of the online median and you get the figures the age selector under your result uses:
| Age group | Expected median on an online test |
|---|---|
| 18 to 24 | about 271 ms |
| 25 to 34 | about 276 ms |
| 35 to 44 | about 281 ms |
| 45 to 54 | about 287 ms |
| 55 to 64 | about 292 ms |
| 65 and over | about 298 ms, and past the range the study covered |
Two cautions. These are expectations built from one slope and one median, not measurements of each group, so treat a 10 ms difference as nothing. And there is no row for under 18 on purpose: reaction time is still getting faster through the teens, and I haven't found a study with numbers solid enough to print.
The practical reading is that age moves the score less than equipment does. Forty years cost about 22 ms in that study. An office monitor and a power-saving wireless mouse can cost that much in one sitting.
Sources: Woods et al., Factors influencing the latency of simple reaction time, Frontiers in Human Neuroscience, 2015; Kosinski, A literature review on reaction time.
How the clock works here
Most reaction tests start the clock when the script decides to turn the box green. The screen shows that change up to a full frame later, 16.7 ms on a 60 Hz display, and that wait lands in your score. This page starts the clock on the frame that carries the green to the screen, so the wait isn't counted. It also reads the time stamp the browser attaches to your click or key press, not the moment the script got around to handling it.
The line under the test shows the refresh rate the page measured on your screen. What's still inside your number and can't be removed from a web page: the monitor's own processing time, anything from a few milliseconds to 20 or more; a mouse polled at 125 Hz, up to 8 ms; wireless and Bluetooth links, which vary a lot. The same person scores 20 to 40 ms better on a gaming PC than on a school laptop. Compare your scores across days on the same machine, not against someone else's setup.
On a phone, add the touchscreen: it samples touches at its own rate and the system adds its own delay before the page hears about a tap. Phone scores usually come out slower than the same person's desktop scores.
Getting a faster score, honestly
- Rest your finger on the button or key with the slack already taken up. Most of the easy gain is in not travelling.
- Use a wired mouse or a keyboard key. Space on a wired keyboard is usually the fastest input most people own.
- Watch the whole box, not its centre, and don't try to predict the change. The pause is random between 1.5 and 4.5 seconds for that reason.
- Practice helps for a few days and then flattens. The history chart under the test shows your last 20 results, so you can see when you've hit your floor.
- Sleep. Reaction tests are what sleep researchers use to measure tiredness, and for a reason.
Questions people ask
What is a good reaction time?
On an ordinary computer, 200 to 250 ms is good and under 200 is fast. Under 180 on a web test usually needs both quick hands and quick hardware.
Is 200 ms good?
Yes. On the estimate used here, a 200 ms median is faster than about 87% of people who take online reaction tests.
Why do I get different results on different sites?
Different sites start and stop the clock at different moments, and some average where this one takes the median. A 15 to 25 ms gap between two honest tests on the same machine is normal.
Can reaction time be trained?
A little. The part that improves is mostly readiness and technique: where the finger rests, how you watch. The nerve and muscle delays underneath don't change much with practice.