Last key pressed
--
- code
- --
- keyCode
- --
- location
- --
Pick the layout that matches your board, then press each key. Keys turn amber while held and green once tested; the progress bar tells you when you have covered them all, and the readout shows the code every press sends. The keyboard latency test runs entirely in your browser, so nothing you press is stored or sent anywhere.
0 / 0 keys tested (0%)
--
none
Total presses
0
Use this keyboard test to test your keyboard in seconds. Open it in any browser, press any key, and see whether your computer registers it, with no drivers or software to install on Windows. This free keyboard tester shows which keys work and which hardware faults need fixing.
This online keyboard tester lights up each press, so a full check takes about a minute. Use the keyboard tester online without signing up, then work through the steps below.
Each press lights up the matching tile on the on-screen virtual keyboard, and a key that works will turn green. Grey tiles are untested, the tile you are holding shows as pressed, and a released key stays tested. A panel also lists the key codes it detects. In my example run on a 104-key board, 101 tiles ended green and 3 stayed untested.
The on-screen board follows an English keyboard in the American keyboard layout (QWERTY), but it reads presses, not positions, so any layout works. The numpad, function keys and modifier keys each get a tile, and the mouse and a game controller have separate checks. Compact boards are covered too: 60%, 75%, TKL (tenkeyless) and full-size. A Mac or Linux machine behaves the same.
A good key tester does more than light tiles. Run a key test whenever typing feels off, and watch performance across the checks below. Hardware managers at an office can clear a stack of laptop keyboard units quickly this way.
A dead key never turns green. An unresponsive key lights late, and stuck keys stay lit after release. Faulty keys like these are the defects to log before you decide on a repair.
Ghosting happens when keys held together fail to register. Press several keys simultaneously: n-key rollover (NKRO) boards register all of them, which matters for rhythm games and multiplayer. In my example, a USB board held 8 keys but registered only 6.
Key chatter is one press that types twice. The test flags each repeat as a double click, usually a worn switch or a debounce time set too low.
Input latency is the delay between a press and the page seeing it, and scheduling in the operating system adds some. The key tester shows when each press registers, and the polling rate sets the floor: \(t = 1000 \div f\), so a 1000Hz board reports every 1ms and a 125Hz one every 8ms when you test your keyboard. Lower input lag and a quicker response speed matter most to gamers, and to gaming above all.
The check runs locally on your device, so your privacy stays intact and the keys you press never leave the page.
I bought a used 87-key tenkeyless board for $38.50 and had a 14-day return window, so the question was simple: keep it or send it back. I opened this online keyboard tester before unpacking the receipt.
First I pressed every key once. All 87 tiles turned green, so there were no dead keys. Then I held W, A, S, D, Space and Shift together, then added Ctrl and E. The page showed 6 of 8 keys registering. That matches the six-key limit of the basic USB keyboard protocol, so it was a limit of the board's mode, not a fault, and I only care about it for fast game inputs.
The real problem appeared on the letter D. I tapped it 60 times at an even pace, and the tester logged 7 extra repeats. That is 7 ÷ 60 = 11.7% chatter, far above anything I would tolerate while typing.
| Run | Taps on D | Extra repeats | Chatter rate |
|---|---|---|---|
| As received | 60 | 7 | 11.7% |
| After compressed air and 90% isopropyl alcohol | 60 | 1 | 1.7% |
Cleaning moved the number from 7 repeats to 1, so I kept the board. What stands out is that one repeat in 60 is still not zero, and the number to watch is the repeat count on that single key, not the overall green screen. A board can pass the coverage check and still chatter.
The edge case: a clean pass on a single run proves little for chatter, because repeats are intermittent. I repeat the 60-tap run twice, and if the rate climbs back above about 3%, the switch is wearing out and I would swap it, since the board is not hot-swap and a return is cheaper than desoldering. Browser input also adds scheduling jitter of a few milliseconds, so I never judge a board on latency from one reading.
When a key fails, work from the cheapest fix to the costliest. Troubleshooting starts with dust and ends at the repair shop.
Dust is the usual culprit. Tip the board upside down, blast compressed air around each keycap, and dab isopropyl alcohol on sticky switches after a spill. Regular cleaning and maintenance keep faults from returning.
Reseat the USB cable, try another port, and test a known-good board in the same socket. A wireless keyboard needs fresh batteries, and Bluetooth links can drop, so a wire is the cleaner check.
If the wrong character types, open the keyboard settings and pick the right language. Then update the firmware and driver software from the maker. Afterwards, press the affected key again to confirm it registers. If nothing helps, a new keyboard is the cheaper fix.
A mechanical keyboard fails differently from a membrane keyboard: the tester shows one dead tile on the first, and often a whole dead row on the second.
A failing part shows as one dead or chattering tile, whether the board uses linear, tactile or clicky switches. Hot-swap boards let you replace just that part, and hot-swapping beats a full rebuild. Typing feel is no guide, since a part that feels fine can still double-register.
Check a gaming keyboard for dropped presses by holding W, A, S, D and Space, with its backlighting on in case a lighting mode disables keys. Check an Apple keyboard on a Mac for its special keys, and check again after you swap keycaps or add a wrist rest, since a misseated keycap can stop a tile lighting.
Open the tester in your browser and press each key. The matching key on the on-screen keyboard changes colour as soon as your computer registers it, so a key that never changes is dead or unresponsive.
A browser-based keyboard tester reads key presses locally and does not need drivers or software. Pick one that states it does not record or upload what you type.
Ghosting happens when you hold several keys at once and some of them fail to register. Press combinations such as W, A, S, D and Space together to see whether every key lights up.
Warning signs are keys that never light up, keys that type twice from one press (chatter), sticky keys, and wrong characters. Run a full pass and note every key that stays untested.
They are layout sizes. A 100% board is full size with a numpad, TKL drops the numpad, and the 75%, 65% and 60% layouts progressively remove the function row, navigation keys and arrows.
Yes. It works with any keyboard type your browser can read, including mechanical, membrane and laptop keyboards, on Windows, Mac and Linux.
Browsers add operating-system scheduling and screen refresh delay to every reading, so a browser measurement is slightly higher than the keyboard's own delay.