Play Reaction Test

Screen · mouse · browser

Input Lag Test

Four checks of the delay your setup adds, measured by your browser. Refresh rate starts on its own; the others need a few moves or presses.

Screen refresh

–

Counting frames for 3 seconds. Keep this tab in front.

Mouse polling rate (estimate)

–

Move your mouse in fast circles here for 3 seconds

Counts every pointer update the browser reports, including coalesced ones. Browsers cap this, so a 4,000 Hz mouse can read lower. Touchscreens don't apply.

Browser event delay

–

Click, tap, or focus here and press keys

The time from the moment the browser stamped your press to the moment our code ran, median of your last 30 presses. On an idle page it's usually under 2 ms; a busy page shows here. Held keys are ignored, so auto-repeat can't inflate it.

Controller (best effort)

–

Connect a controller and press a button to wake it up. We show the time from the controller's last report to the next frame, so it's rounded to your frame interval.

What this can and can't measure

A browser can't see photons leaving your screen. Real end-to-end lag, from the moment you click to the moment the pixels change, needs a high-speed camera or a photodiode stuck to the display. What a browser can measure is its own part of the chain:

  • Refresh rate and dropped frames, by timing requestAnimationFrame for three seconds. A frame that takes more than 1.5 times the usual interval counts as dropped.
  • Mouse polling, by counting pointer updates while you move, including the coalesced ones the browser groups together. It's an estimate, capped by the browser.
  • Event delay, the gap between the hardware timestamp of your press (event.timeStamp) and the moment our code ran. On an idle page that's tiny; a busy page shows up here.
  • Controller, using the Gamepad API: time from the controller's last report to the next frame, rounded to your frame interval.

Display lag explained

Display lag has three parts. The wait for the next refresh: up to 16.7 ms at 60 Hz, 6.9 ms at 144 Hz. The monitor's processing: scaling, overdrive and picture modes, often worst on TVs. And the pixels changing colour. In Woods et al. (2015) the lab display added 11.0 ms. Human Benchmark notes that some modern TVs add as much as 150 ms, which is why TV game modes exist.

Browsers add their own timing error on top. Anwyl-Irvine et al. (2021) measured mean display timing errors of 11.50 ms in Chrome, 18.72 ms in Edge, 22.58 ms in Firefox and 30.02 ms in Safari on their test machines.

Mouse and controller polling

A mouse reports its position and buttons at a fixed rate: 125 Hz is every 8 ms, 1,000 Hz every 1 ms. Higher polling means a click is reported sooner on average. Keyboards vary much more: the same study measured keyboard response lag averaging around 80 ms on their rigs, up to 100 ms on some set-ups. Wireless and Bluetooth devices add radio delay, sometimes a lot.

How lag changes your reaction test

Every millisecond of lag lands in your reaction time. Our reaction time test starts the clock on the frame where the green is drawn and stops at your press's timestamp, which removes our own code from the number, but the screen and input device are still in it. That's why each result names your device, input and refresh rate, and why we explain our method rather than quietly subtracting a guess. Fix what you can, then compare scores on the same setup. Gamers: see gamer reaction time.

Sources

  1. MDN, requestAnimationFrame()
  2. MDN, Event.timeStamp
  3. Anwyl-Irvine et al. (2021), Realistic precision and accuracy of online experiment platforms, web browsers, and devices, Behavior Research Methods 53:1407-1425
  4. Woods et al. (2015), Factors influencing the latency of simple reaction time, Frontiers in Human Neuroscience 9:131
  5. Human Benchmark, reaction time statistics
  6. Wikipedia, Display lag
  7. Wikipedia, Input lag

FAQ

Input lag questions

How can I test input lag on my PC?

For the parts a browser can see, use the checks on this page: refresh rate and dropped frames, mouse polling and browser event delay. For true end-to-end lag, from click to light on the screen, you need a high-speed camera or a light sensor on the display.

Is 20ms input lag noticeable?

Whether you feel it depends on you and the game. It is more than a whole frame at 60 Hz (16.7 ms), and every millisecond of it lands in your reaction time, so it shows up in your score either way.

What is a good latency for gaming?

Lower is better and every part adds up: a high refresh rate screen in its fastest mode, a wired mouse or keyboard, and a browser or game that isn't dropping frames. Measure what you can here and compare after each change.

Keep playing

More reaction tests