Reaction time test

Each try is armed by a press on the pad, then a random wait passes before the signal, so its moment cannot be learnt. The clock starts with the screen frame that draws the visual stimulus and stops when your finger, mouse button or key goes down. Pressing before the signal, or too soon after it to be a genuine reaction, is a false start, and that attempt is run again.

The wait is drawn at random between 1.5 s and 4.5 s, in steps of 10 ms, for every try and again after every early press. Anything less than 100 ms after the signal is treated the same way. With no press within 2 s, the try is a miss and is run again. A run measures five or ten reactions, five by default.

On a phone the pad takes a tap, and on a computer a click. Space starts a run once the pad has focus. During a run any key counts as a press, apart from Tab, Esc and keys such as Shift, and Esc stops the run.

Reaction time is the interval between a signal and the response to it, here given in milliseconds, thousandths of a second. It is a different skill from pressing many times in a row, which the click speed test counts in clicks per second.

Each mode changes the signal while the clock and the scoring stay the same. Three of them measure simple reaction time, one known signal answered with one response, and the fourth adds a decision before every press.

• Red to green: the pad is red during the wait and turns green at the signal, the classic simple task.
• Start lights: five red lights come on one by one, the first 0.7 s after the try is armed and then one every 1 s. All of them go out after a random hold of 0.2 s to 3 s, and lights out is the signal. That hold range is a choice made for this game.
• Go or stop: a green circle means press and a red cross means hold back. The game deals 0.4 stop signals for every go signal: 2 in a run of five and 4 in a run of ten. They are shuffled, but never first and never two in a row.
• Sound: the screen does not change, and the signal is a short beep.

Psychologists call the green and red mode a go/no-go task. A stop signal stays on screen for 1 s. Holding back counts as a stop signal held, while a press counts against you and the run carries on.

In a choice reaction time task, several signals each have their own response, so every press needs a decision about which one. The Stroop test works that way, with an ink colour to pick for every word.

A good reaction time, by the bands used here, is an average below 250 ms, and below 200 ms counts as exceptional. The typical band runs from 250 ms to 350 ms, steady from 350 ms to 450 ms, and slower than typical from 450 ms. These bands apply to every mode without stop signals; the go/no-go bands sit 50 ms higher, since each press waits on a decision.

The scale is this site's own, placed around a large published web sample. Passell and colleagues timed 30,747 people on a red square turning green and found a median of 304 ms, with a standard deviation of 69 ms. Like this game, that figure includes the delay of each person's own device, so laboratory figures taken with fast equipment run lower.

Below 200 ms is exceptional in three modes, and below 250 ms for go/no-go.

Published figures cover a wide span: a laboratory mean of 231 ms with a gaming mouse, the web study's 304 ms, and 539 ms for matching cards.

UK Biobank's figure comes from a slower task with middle-aged and older adults, who pressed a button when two cards matched, a decision on every round. Its showcase treats times under 50 ms as anticipation and ignores those over 2,000 ms, so the figure describes that task rather than this one.

The game gives no percentile, because it collects nobody's results to rank you against. The comparison it keeps is your own: a personal best for each mode, stored in this browser.

The average reaction time is the headline figure: the measured reactions added up and divided by their number. In an illustrative run of five reactions, 312 ms, 268 ms, 295 ms, 241 ms and 289 ms, the total is 1,405 ms. The average is 281 ms, in the typical band.

The median is the middle time once the reactions are sorted, 289 ms here, so one slow outlier barely moves it. The fastest try took 241 ms and the slowest 312 ms. The spread, a sample standard deviation, is 27 ms. It measures consistency: a small spread means the tries were alike.

Early presses and misses are counted and shown, but they never enter the times, since each of those tries is run again. When a run has stop signals, the result also shows those held, such as 2 of 2. A press on a stop signal is the classic no-go error, and such a run becomes your best only when every stop signal was held.

Each reaction is listed with the input that made it: mouse, touch, pen or key. "Copy result" copies a line of text with a link, and the link carries only the mode and the number of reactions, such as ?mode=lights&tries=10.

The best average for each mode and your last ten runs stay in this browser's local storage, and "Clear my results" deletes them. Nothing is sent anywhere.

In the sound mode the screen stays the same, and the signal is an auditory stimulus: a 1,000 Hz beep lasting 150 ms. Laboratory studies usually find reactions to a sound a little faster than to a light. Jain and colleagues saw the same in 120 medical students, who answered a beep faster than a light.

A browser plays sound through its own audio path, which adds a delay of its own. The game times the beep from the moment the browser reports it reaching the speakers, using the output timestamp of the Web Audio API. The W3C standard describes output latency as an estimate that depends on the platform and the audio device connected.

Bluetooth headphones and other wireless ones can add a delay the browser may not fully report, so a result on them can read slower than the reaction itself. That is why the sound bands match the colour bands rather than sitting lower. The "Play a test beep" button checks the volume first, since a phone on silent may stay quiet.

Every device adds time in two places. The screen shows the signal a little after it is drawn, and the mouse, keyboard or touchscreen reports the press a little after it happens. Reimers and Stewart measured 19 Windows systems and found web response times overestimated by 30 ms to 100 ms, depending on the system.

Within one system the spread was generally under 10 ms, which is why tries on the same device compare well. Pronk and colleagues pressed with a robot, one device for each system. The overestimate was about 58 ms on iOS Safari, 69 ms on Windows Chrome, 70 ms on Android Chrome and 133 ms on macOS Safari.

In the laboratory work of Woods and colleagues, the LCD monitor added 11.0 ms and the gaming mouse 6.8 ms. That input lag and display lag accounts for most of the gap between their mean of 231 ms and the 213 ms left once the hardware delay was taken out.

The refresh rate sets how often the screen draws a new picture: at 60 Hz, one frame every 16.7 ms. The game starts the clock at the start of the frame that draws the signal and stops it at the press's own timestamp. A mouse's polling rate, how often it reports to the computer, adds its own small share.

A touchscreen, a trackpad, a wired mouse and a wireless one each report presses in their own way, so changing device moves the figure. A faster result on a new device can be the hardware rather than the person.

At 30 mph a car travels 3.8 m (12 ft) during the example average of 281 ms, and 8.8 m (29 ft) at 70 mph. Your result shows the same two figures for your own average, at the built-up and motorway limits.

The Highway Code's typical stopping distances start with a thinking distance at each speed in mph. The Code states no reaction time itself, but those distances work out at about 0.67 s, which is 9.0 m at 30 mph. AQA's GCSE Physics specification defines thinking distance as the distance the vehicle travels during the driver's reaction time.

At 30 mph, a reaction of 200 ms covers 2.7 m, while the 670 ms worked out from the Code covers 9.0 m.

A press on a screen is not a reaction in traffic. A driver has to notice the hazard, decide to brake and move a foot to the pedal before the brakes begin to work. The Code adds that thinking distance depends on attention, and its diagram gives each stopping distance in car lengths too.

Braking distance comes on top, and the stopping distance calculator adds it for a chosen speed and road condition. Rule 126 asks drivers on fast roads for at least a two-second gap to the vehicle in front, often called the two-second rule. On wet roads the gap is at least doubled.

The DVSA's hazard perception test, part of the theory test, measures something else. It shows 14 video clips and gives up to 5 points for each developing hazard, scored by how soon you spot it starting. The pass mark for cars is 44 out of 75.

That test rewards anticipation, not a reaction in milliseconds, and clicking continuously or in a pattern scores nothing. No UK driving licence uses a timed reaction test, and this game neither copies nor prepares for any part of the theory test.

Two of the game's rules come from racing, where a start ahead of the signal is penalised. The start lights mode follows the five red lights of a motor race start. The false start line comes from sprinting, where athletes leave starting blocks linked to a start information system.

The FIA's 2026 Formula 1 Sporting Regulations, in the issue of 1 October 2026, switch the five red lights on one by one at intervals of 1 s. The time between the fifth light and all lights going out is at the sole discretion of the permanent starter, so the rules set no range. The game's hold of 0.2 s to 3 s is its own.

Under the same rules a car must be stationary after the third red light until the lights go out, and a transponder detects when it first moves. Moving too early is the offence race reports often call a jump start.

World Athletics draws the sprint line in its Technical Rules of 1 July 2026. Its start information system flags a possible false start when the reaction time is less than 0.100 s. An athlete responsible for one is disqualified, except in combined events. The game borrows the same line, so a press under 100 ms after the signal is too soon to count.

That line is a rule for certified start systems, not a human limit. Lipps and colleagues analysed 425 sprinters at the Beijing Olympics: the mean of each athlete's fastest reaction was 166 ms for men and 189 ms for women. Their analysis estimated that the quickest sprinters can react in as little as 109 ms (men) and 121 ms (women).

Fatigue, alcohol, drugs and distraction can all lengthen a reaction. AQA's GCSE Physics specification names tiredness, drugs and alcohol as factors in a driver's reaction time, and adds that distractions may affect the ability to react. It gives typical values from 0.2 s to 0.9 s.

The Highway Code makes the same points for drivers. Rule 95 says alcohol will reduce co-ordination and slow down reactions. Rule 91 warns that driving while tired greatly increases the risk of a collision. Rule 148 adds that safe driving needs concentration.

Readiness counts as well. Woods and colleagues measured reactions roughly 15% slower after the shortest waits before the signal, when people had less time to prepare. The waits here are random, so some tries catch you less ready than others.

Repeating runs makes the task familiar, and a change smaller than your usual spread may be chance rather than practice. GCSE science includes a practical on how one factor changes human reaction time. An average of ten reactions per condition varies less from run to run than one of five.