Click the button labelled Click target, not the surrounding panel. The next input is ignored until the two animation-frame callbacks finish and the target moves.
Part of the Mouse test. Run the full mouse test first for a complete device check.
The target does not advance Results vary widely The value exceeds another utility
Mouse latency test
Measure click-to-paint delay over repeated targets and report median, tail and uncertainty values.
Also known as: mouse latency test, click latency test, mouse input lag test, click delay test, mouse delay test, mouse response time test. Free, online, and no download required.
No permission neededNothing uploaded
Press the target when it appears to begin collecting samples.
01
What the mouse latency test checks
| Check | How it is measured | Fails when |
|---|---|---|
| Input timestamp | The pointer event timestamp anchors each sample. | The browser supplies an invalid or repeated timestamp. |
| Paint completion | Two animation frames bound the next visible paint opportunity. | A sample never reaches the paint callback. |
| Distribution | Median, 95th percentile and median absolute deviation summarize the set. | Fewer than ten valid samples arrive. |
How to run the mouse latency test
- Keep the tab visible and click only the moving Click target button. Wait for it to shift before the next click.
- Collect at least twenty completed samples. Read the median, p95 and timing uncertainty instead of selecting one fastest value.
- Repeat without moving the window or changing display mode. Compare one browser, connection or power condition at a time.
If something failed
Keep the window still and foreground, close heavy background work and collect at least twenty targets. Compare median, p95 and uncertainty across complete runs.
Check what the other utility measures. This mouse latency test includes browser dispatch and two animation-frame callbacks; external hardware reviews often use different endpoints and equipment.
02
How to read the result
What is a good mouse response time?
There is no universal good value for this page. The result is click-to-paint timing inside the current browser pipeline, not direct mouse response time or sensor-to-photon hardware latency.
Compare complete runs on the same browser and display mode. Did median, p95 and timing uncertainty move together after one condition changed? That is stronger evidence than one fastest sample.
Use the input latency guide to separate device reporting, browser work, display frames and network delay. For movement-event frequency instead of click-to-paint timing, use the mouse polling rate test. This test covers only the browser-observed portion defined below.
How the mouse latency test works
A pointerdown on the Click target supplies the event timestamp. The page schedules one requestAnimationFrame callback and then a second; the second callback records performance.now minus the event timestamp and moves the target.
Why two frames? The first callback runs before a paint opportunity, while the following callback bounds the next visual update used by this test. That method includes browser and display scheduling; it does not identify the physical mouse delay.
The page ignores another latency click while those callbacks are pending. Click only after the target shifts so each accepted action creates one completed sample.
What the result can establish
Collect at least twenty targets with the tab visible and the window still. The current verdict requires enough valid samples; stopping early leaves the result incomplete.
Median, also labelled p50, represents the centre of the recorded click delay test samples. p95 shows the slower tail, and median absolute deviation becomes timing uncertainty. Read all three before comparing runs.
The fastest sample is not the result. A single callback can land differently within a display frame. Compare repeated distributions under the same browser, screen and power state.
Measurement limits
Mouse latency test includes event dispatch, JavaScript, style, layout, rendering and compositor scheduling around this page's target. It does not isolate USB, wireless transport, switch or panel latency.
A wireless and wired comparison may differ, but browser timing alone cannot assign the difference to the radio. Keep every other condition fixed and repeat before drawing a narrower conclusion.
Keyboard latency uses a KeyA event and a separate page. Mouse and keyboard samples have different input paths, so use each tool for its stated target rather than ranking devices across unlike tests.
03
Mouse latency test questions
What does this mouse input lag test actually measure?
The mouse input lag test measures from the accepted pointer event timestamp through two animation-frame callbacks. It includes this page's browser rendering path, not direct hardware latency.
Why is my click latency test result higher than a review's?
A review may use external sensor-to-photon equipment and different endpoints. This click latency test includes browser and compositor scheduling, so compare it with repeated runs of the same test instead.
How many samples should a click delay test collect?
Collect at least twenty completed Click target samples with the tab visible. Read median, p95 and uncertainty; one click delay test sample is too sensitive to frame timing.
What is a good response time for a mouse?
There is no universal good response time for this measurement. Use a controlled baseline on the same browser and display, then compare one changed condition.
Does a wireless mouse add input delay?
Wireless can change the end-to-end path, but this page cannot isolate radio delay. Compare wired and wireless complete runs while keeping browser, display and background load fixed.
How should I repeat the mouse latency test fairly?
Keep the tab foreground, click only after the target moves and collect the same sample count. Do not move the window or change display mode between runs.
What can the mouse latency test not prove?
The mouse latency test cannot report raw sensor, USB, radio or panel delay. It measures the defined click-to-paint path in this page.