Skip to main content

Reaction time test

Measure how fast you react to visual, audio, and aim stimuli — timed with performance.now() precision.

Free & unlimitedWorks offline
Mode
Attempts
Press Start test to begin
Click when the screen turns green

Average human reaction time is roughly 250 ms for visual stimuli and 170 ms for audio. Pro gamers typically average 150–200 ms.

Consistency (low standard deviation) matters more than a single fast attempt. Timing uses performance.now() for sub-millisecond precision.

All processing happens in your browser. No data is sent to any server.

This reaction time test runs three different stimuli through the same timer. Visual waits a random 1 to 5 seconds and then changes colour. Audio plays a 150 ms sine tone at 800 Hz through Web Audio with no visual cue at all. Aim spawns a 30-pixel-radius target somewhere inside the pad and scores both how fast you hit it and how close to its centre you landed. Every attempt is timed with performance.now, clicking during the wait is recorded as a false start rather than a fast time, and at the end of the run you get average, best, worst and a standard deviation that measures how consistent you were.

Key facts about Reaction time test

Key facts about Reaction time test
ModesVisual colour change, Audio 800 Hz tone, Aim click target
Wait before the stimulusrandom, between 1.0 and 5.0 seconds, so the moment cannot be anticipated
Timing sourceperformance.now(), read at the stimulus and again on your click, rounded to whole milliseconds
Audio cuea sine OscillatorNode at 800 Hz, 0.3 gain, lasting 150 ms
Aim target30-pixel radius, placed at random inside the pad, fully clear of the edges
Aim accuracydistance from the target centre expressed as a percentage of its radius
False starta click during the wait phase ends the attempt with no time and is excluded from the average
Attempts per run3, 5, 10 or 15 — five by default
Consistencysample standard deviation across the valid attempts of the run
Rating bandsunder 150 Incredible, under 200 Excellent, under 250 Good, under 300 Average, under 350 Below average, above that Slow
Session historyeach completed run is kept with its mode, average and best while the page stays open
ExportJSON with every attempt, the four statistics, the rating and the session history

What happens to your file

Everything runs in this tab. The random delay comes from Math.random in your browser, the timestamps come from performance.now, the beep is synthesised locally by a Web Audio oscillator rather than downloaded, and your results are held in React state. No score is uploaded, there is no leaderboard, and nothing is written to local storage — reloading the page clears the run and the session history. The JSON export is built in memory as a Blob and downloaded by your own browser, so even that never crosses a network.

About this tool

  1. 1

    Choose a mode

    Visual is the standard test most people mean. Audio removes the display entirely from the chain. Aim adds the physical movement of the mouse, so it measures something different rather than something harder.

  2. 2

    Set how many attempts

    Three is enough for a quick check, five is the default, and ten or fifteen gives a standard deviation worth trusting. Longer runs also show fatigue, which is itself informative.

  3. 3

    Wait for the stimulus

    The delay is random between one and five seconds precisely so it cannot be timed. Clicking early is recorded as a false start with no time, and it does not pollute your average.

  4. 4

    React, do not predict

    In visual mode watch the pad, not the clock. In audio mode close your eyes if it helps — nothing on screen changes. In aim mode keep the cursor near the centre so travel distance is comparable between attempts.

  5. 5

    Click through to the next round

    After each attempt the result stays on screen until you click again, which starts the next wait. The run ends automatically when you reach the attempt count.

  6. 6

    Read the spread, not just the average

    Average with a small standard deviation means genuinely fast. A fast average with a large deviation usually means a couple of lucky anticipations mixed with slow attempts.

  7. 7

    Export or repeat

    Download the JSON if you are tracking change over time, or run the same mode again — the session history keeps every completed run for as long as the page is open.

Specs & compatibility
Browser supportall modern browsers; audio mode needs Web Audio, which Safari exposes as webkitAudioContext
Timer resolutionperformance.now() is deliberately coarsened by browsers, typically to 0.1 ms or worse — fine at this scale
Display latencyincluded in visual mode: a 60 Hz screen adds up to 16.7 ms that is not your reflexes
Input latencyincluded in every mode: mouse polling, USB scheduling and the browser event loop all sit inside the number
Audio latencyincluded in audio mode: speaker and driver delay is part of the measured time
Touch deviceswork in all three modes; a tap registers the same as a click, though touch adds its own delay
What the score is nota clinical measurement — it is your reflexes plus your whole hardware chain
Account or installnone — no signup, no app
  • Take one throwaway run before the one that counts; first attempts are consistently slower while you learn the pad.
  • Audio results usually come in 20 to 40 ms faster than visual, because the auditory pathway is shorter — that is expected, not a fluke.
  • Aim results are much slower than visual because they include moving the mouse; compare aim to aim, never aim to visual.
  • A 60 Hz monitor can add up to 16.7 ms to every visual attempt. Running the test on a 144 Hz screen will look like an improvement in reflexes that is really an improvement in hardware.
  • Wireless mice and Bluetooth peripherals add a few milliseconds; if you compare setups, note which one you used.
  • Test at the same time of day. Sleep, caffeine and fatigue move reaction time more than any equipment change.
  • A false start is information: repeated false starts mean you are guessing the delay rather than reacting to the stimulus.
  • If you want a clean comparison between two computers, use audio mode, which removes display latency from the chain.
  • Visual, audio and aim modes sharing one timer
  • Random 1 to 5 second delay with false-start detection
  • 3, 5, 10 or 15 attempts per run
  • Average, best, worst and standard deviation
  • Aim accuracy scored against the target centre
  • Rating from Incredible to Slow
  • Session history and JSON export
  • Getting a baseline reaction time before tuning a gaming setup.
  • Comparing reflexes on two machines, or before and after a monitor upgrade.
  • Checking whether caffeine, sleep or a warm-up actually changes your numbers.
  • Practising aim and reaction separately, since the tool measures them as different things.
  • Settling an argument between friends on identical hardware.
  • Tracking consistency over a week using the standard deviation rather than the best single attempt.
For simple visual reaction, most adults land between 200 and 300 ms, and the tool calls anything under 250 ms good and under 200 ms excellent. Trained players often average 150 to 200 ms. Under 150 ms is possible but rare, and when it appears alongside a large standard deviation it usually means anticipation rather than reaction. Audio reaction is naturally faster than visual by roughly 20 to 40 ms. Aim mode is much slower for everyone because it includes moving the mouse.
Because measured reaction time includes your whole chain, and the biggest variable is the display. A 60 Hz monitor can add up to 16.7 ms to a visual test before your eyes are involved, and the browser, the mouse and the operating system add more. Different sites also start their timer at slightly different moments. Comparing your own results across days on the same hardware is meaningful; comparing a number from this page against a number from another site is not.
It measures how much your attempts varied around their own average. A run of 210, 215, 208 and 212 ms has a small deviation and describes a genuinely stable reaction. A run of 160, 290, 180 and 300 ms can produce a similar average with a large deviation, which usually means you anticipated on some attempts and reacted late on others. For tracking improvement, a falling deviation is often a better sign than a falling average.
Because it arrived during the random wait, before the stimulus appeared. The delay is drawn fresh between one and five seconds for every attempt precisely so it cannot be learned, and a click in that window is a guess rather than a reaction. False starts are recorded as attempts but carry no time and are excluded from the average, best, worst and deviation, so one impatient click does not distort the run.
No, and it should not be compared with it. Visual mode measures how long it takes you to notice a change and press a button that is already under your finger. Aim mode adds target acquisition: seeing where the target appeared, moving the mouse there, and clicking accurately. That movement dominates the time, which is why aim scores are several hundred milliseconds slower. The accuracy percentage next to each aim attempt shows how close to the centre you landed, which is the part that rewards practice.
No. Completed runs are kept in memory as a session history so you can compare runs within one visit, and that is all. Nothing is written to local storage or cookies, nothing is sent to a server, and there is no leaderboard or account. Reloading the page wipes the history. If you want to keep results, use the export button, which builds a JSON file in your browser containing every attempt, the statistics and the session history, and downloads it locally.

Related tools

View all

Part of Hardware testers that run in your browser

Updated

We use anonymous analytics to improve ToolChamp. No personal data is stored or sold. Privacy Policy