Mouse DPI Test
Pick a distance, hold the left button and drag steadily across the lane while the mouse physically travels the same distance on your desk — release to see the estimated DPI, the nearest common step, and the median across runs.
Hold the left button anywhere in this lane and drag straight to the right, then release.
For a true reading, OS pointer speed must be at the raw 1x position (Windows: slider centered, "Enhance pointer precision" off).
How to read it
- DPI = on-screen pixels divided by physical inches moved. The tool shows the estimate and the nearest common step (400/800/1600…).
- Keep the drag straight and steady — jerky or diagonal motion inflates the pixel count. Three runs, take the median.
- macOS and trackpads apply acceleration — the reading reflects system-translated speed, not hardware DPI.
- Far off the advertised figure? Check the DPI cycle button on the mouse and the driver software sensitivity first.
Frequently asked questions
How can a web page measure DPI?
DPI is pixels moved per physical inch. The browser counts the pixels you drag across the screen while you measure the physical travel with a ruler — divide one by the other for the estimate. Prerequisite: OS pointer speed at raw 1x — on Windows keep the slider centered and 'Enhance pointer precision' off, or the result scales with the multiplier.
My result is higher/lower than the advertised DPI — why?
Confirm the OS pointer speed is centered, pointer acceleration is off and driver sensitivity is 1x. Still off? The driver's DPI step may differ from what you expect, or the DPI switch button on the mouse was pressed — cycle it and retest.
Does it work on macOS or a trackpad?
macOS applies a pointer acceleration curve by default, so you measure the translated speed, not hardware DPI — treat it as indicative only, same for trackpads. For a true DPI reading use a wired or 2.4GHz mouse on default Windows settings.
Is landing between two DPI steps normal?
Yes. Hand-drawn lines aren't perfectly straight and ruler readings carry millimeter error — ±5% is expected. Run the test a few times, take the median, and read off the nearest common step (400/800/1600/3200).