What this test measures
The panel starts red with the words "Wait for green." After a random wait of 1.5 to 7 seconds (a deliberately wide window, so the change cannot be predicted) it turns green. Your job is to click or tap the instant you see the change. The time between the color change and your click is your reaction time: how long it takes you to spot one expected signal and answer it.
You get five rounds. Your score is the median of the five, and your single best round is shown alongside it. If you click while the panel is still red, that round is voided as "too soon" and does not count toward anything. The delay is randomized precisely so you cannot time it; guessing is the thing being filtered out.
One quirk: if the browser tab is hidden, the test refuses to run. Browsers throttle timers in background tabs to save battery, which would corrupt the measurement, so the test simply waits until the tab is visible again.
The science
The comparison figures on this page come from published research, not from a database of other visitors' scores. The anchor study is Woods and colleagues (2015), who measured simple reaction time in 1,469 adults aged 18 to 65. Their participants averaged 231 ms, or about 213 ms once the researchers corrected for the 17.8 ms of delay introduced by their own testing hardware. People differed from each other by about 26.8 ms either way. And reaction time crept up with age, by about half a millisecond per year.
Your browser adds its own delay, and it is not small. Anwyl-Irvine and colleagues (2021) measured how much delay browsers and devices add, across four online test platforms. The answer: 71 to 87 ms on average. That lag sits between your finger and the timestamp the page records, and it counts against you.
That delay is exactly why this test shows no percentile. Correcting every visitor with one flat 80 ms figure would flatter fast machines and shortchange slow ones: two people with identical nervous systems would land at different "top N%" purely because of hardware. A browser test is not the same experiment as the lab study, so instead of pinning a percentage on you, the result screen tracks your own trend on your own device, and the research numbers stay here as reference for reading your score yourself.
What counts as a good reaction time
On a simple task like this, a typical adult truly reacts in roughly 200 to 250 ms. In the Woods study, most people fell between about 186 and 240 ms. That is the corrected average of 213 ms, give or take one typical step. A true reaction time under about 185 ms already makes you faster than most adults in that sample.
On this test, though, think in on-screen numbers. With typical browser latency added back, an average adult should expect a displayed median somewhere around 280 to 300 ms. A displayed 250 ms is quick. A displayed 220 ms or lower is rare, and at that level you should suspect your hardware is flattering you a little.
The scores guide lays out the full scale, from under 150 to over 360, with the badge each band earns and what 186, 200, 213, 250 and 300 ms each mean on the screen scale and the lab scale.
Age shifts the picture slowly. At 0.55 ms per year, the average 60-year-old is only about 22 ms slower than the average 20-year-old. That gap is smaller than the round-to-round noise in your own results. A slow score on any one day says more about the day than about you. The 0.55 figure is the laboratory slope; large browser datasets report a steeper one, and the age guide explains why the two disagree.
Why your score jumps around
Run the test twice and you will probably get two different numbers. That is expected. Woods and colleagues found intra-subject variability of roughly 40 ms, meaning one person's individual trials routinely swing that much around their own average. Five rounds will not smooth this out completely, but it is enough to make the median meaningful.
The median is used instead of the average because it shrugs off outliers. One distracted round of 500 ms would drag an average badly but barely moves a median. A lucky anticipation that slips through past the red-click filter does the same in the other direction. The best-round figure is shown because it is satisfying to see, but the median is the honest number.
Hardware adds noise of its own. Screen timing, mouse timing and system load all wobble from moment to moment. That wobble is part of the 71 to 87 ms of delay in the Anwyl-Irvine data. Background tabs are worse than noisy: browsers slow their clocks on purpose. That is why the test refuses to run while the tab is hidden, instead of handing you a broken result.
How to improve it
Honestly, not by much. Simple reaction time is a fairly stable trait, and most of the improvement people see comes from measurement conditions, not from a faster brain. The conditions are still worth controlling:
- Sleep. Reaction-time tasks are a standard laboratory probe for sleep deprivation. If you test while exhausted, you are measuring the exhaustion.
- Caffeine timing. If you are a regular coffee drinker, test after your usual cup, not before it. Consistency matters more than the dose.
- Practice. Expect your first session to be your worst. Early gains mostly reflect learning the task itself, such as where the panel sits and how the click feels, and they flatten quickly.
- Posture and hand position. Rest your finger on the mouse button or hover your thumb just over the screen. Travel distance is dead time.
- Screen refresh rate. A 60 Hz display draws a new frame every 16.7 ms, so the green can appear late. A 144 Hz monitor cuts that wait. This improves your displayed score without changing you at all.
If you are chasing a number, run full five-round sets and compare medians across days instead of celebrating single flukes.
Frequently asked questions
Why was my round voided as "too soon"?
You clicked while the panel was still red. That means you were predicting the change, not reacting to it, and a prediction says nothing about your reaction time. The delay is drawn from a deliberately wide 1.5 to 7 second window to make prediction unprofitable; a voided round is that filter doing its job.
I clicked after the green appeared. Why was it still voided?
Because the time was under 100 ms. The physical journey alone — light into the eye, signal through the brain, command down to the finger muscles — takes more than a tenth of a second, so no click under that line can be an answer to the signal, whoever makes it. Athletics uses the same rule: a sprinter who leaves within 100 ms of the gun is charged with a false start. Anything above that line counts in full, however fast. A machine cannot tell a sharp reaction from a lucky guess, so this test only filters what is physically impossible.
Why is my phone slower than my computer?
Different devices add different amounts of delay. Touchscreens even process your tap through their own circuits before the page hears about it. The Anwyl-Irvine study found delays varying across the browsers, operating systems, and devices they measured. The 80 ms correction is a single average, and your particular phone may sit above it. The devices guide puts the measured phone delay next to the desktop figure and explains which score to trust.
Is 300 ms a bad score?
Displayed on this test, no. Take off the 80 ms of device delay and you are at roughly 220 ms of true reaction time. That is close to the corrected average of 213 ms from the Woods study. It is an ordinary, healthy number.
Why is my score faster here than on other reaction sites?
Different sites stop the clock at different moments. This test records the instant you press (pointerdown), while many others record the instant you release (click). Pressing and releasing takes roughly 60 to 100 ms, so the release-timed sites read that much slower for the exact same tap. The final score differs too: we take the median of 5 rounds, so one fumbled round cannot drag your score, whereas an average-based site mixes that slow round straight in. Neither approach is wrong: the rulers simply start at different points. Compare your own progress within one site instead of comparing numbers across sites.
Why will the test not start in a background tab?
Browsers throttle JavaScript timers in tabs you are not looking at, sometimes severely. Any measurement taken under throttling would be garbage, so the test declines to run at all until the tab is visible.