Testers

Keyboard Tester

Test every key on your keyboard in the browser. See which switches register, catch ghosting, check 100%, 80%, 75%, 65% and 60% boards in ANSI, ISO or AZERTY.

Reads real keydown/keyup events on this page. Tab and Escape always keep their normal behaviour, so you're never stuck here.

Size
Standard
not yet pressedregisteredheld down now
Last key
—press any key
Presses
0
Keys tested
0 / 104
Chattering
None
Largest rollover
0 keys
Best here
0 keys rollover0 keys · 0 presses all-time

Key log

Press any key to start.

Press every key on your keyboard and watch it register in real time. Keys that respond turn cyan; keys that stay pale are the ones worth investigating further. What you type isn't uploaded — the whole test runs inside this tab, and your best sessions are remembered only in this browser's local storage.

How to test a keyboard properly

Work along each row rather than hunting for the one key you already suspect. A dead switch usually has neighbours with the same problem, and a whole row failing together points at a loose ribbon cable or connector, not a single worn switch.

  • Set the two pickers above the board to match your keyboard before you start. The first is the size — 100%, 80% TKL, 75%, 65% or 60% — and the second is the standard, ANSI, ISO or AZERTY. They are separate questions because a board answers both independently; an ISO 75% is an ordinary thing to own.
  • Press every key once, including both Shift keys, both modifier keys, and the full function row — it's easy to skip keys you rarely use and miss a fault sitting right next to a frequently used one.
  • Hold three keys down at once and confirm all three light up together. A key that drops out under load is being blocked by the keyboard's internal switch matrix — this is ghosting, and it's a hardware limit, not a software bug.
  • Tap a suspect key rapidly, twenty times or so, and watch the key log. A key that logs two presses for one physical tap is chattering, usually a sign the switch contact is worn or dirty.
  • If an entire cluster of keys is dead rather than one isolated switch, reseat the keyboard's cable or try the board on a different machine before you conclude the switches themselves have failed.

Ghosting test and key rollover test: what the readout means

The tester listens for the browser's event.code, not the character your operating system would normally produce, so it reports the physical switch you pressed rather than the letter printed on the keycap. That distinction is why the key marked A on an AZERTY board reports as KeyQ — physically, it's the switch position a QWERTY board prints Q on — until you switch the on-screen layout above to match your keyboard's actual print. AZERTY moves only four letters relative to QWERTY, swapping A with Q and Z with W, so those are the keys where the mismatch is obvious; Y, for instance, sits in the same place on both and reports as KeyY either way.

Each key on the board shows one of three states: not yet pressed, held down right now, or registered at some point in this session. The key log records every keystroke with its reported key, its underlying code, and how long the switch stayed down in milliseconds, which is useful for spotting a switch that registers but releases unusually slowly. Session stats summarise the run — total keys pressed, unique keys hit, and any chattering or ghosting flagged along the way — so you can compare one test run against the next without re-reading the whole key log.

Numpad test: checking a full-size board's number pad

Select the 100% size above and the board grows a navigation cluster and a 17-key numpad, which is the only place on this page a numpad can be tested — every smaller size genuinely does not have one, so the keys are not drawn. Work down the pad the same way you work along a row: Num Lock, the four operators, the ten digits and the decimal point, then the numpad's own Enter.

Two results are worth telling apart. If a numpad key lights up here but types nothing where you actually need it, the switch is fine and Num Lock is off — with Num Lock disabled those keys send Home, End, Page Up and the arrows instead of digits. If a numpad key does not light up here at all, the switch or its wiring is the problem. Note also that NumpadEnter and the main Enter are two different physical switches reporting two different codes, which is why a board can have one working and the other dead.

Key log: spotting a chattering key

Copy the key log if you want to keep a record of a test run, share it with a seller before buying a used keyboard, or compare two boards side by side. Because everything is derived from event.code, the log stays consistent across layouts — the underlying reading of the switch never changes, only how it's labelled on screen.

Questions

Why does my AZERTY keyboard show the wrong letter when I press a key?
Because the tester reports which physical switch you pressed, not which character it prints. Browsers expose keyboard input through two separate signals: event.key, the character produced under your current layout, and event.code, the physical position of the switch on the board. This tester reads event.code, so a French AZERTY board's A key — which sits where a QWERTY board prints Q — reports as KeyQ regardless of what's on the keycap. AZERTY moves four letters relative to QWERTY: A and Q swap places, as do Z and W, which is why those four are the keys that look wrong. Select AZERTY from the layout switcher and the on-screen labels match your keycaps again; the underlying code never changes, only how it's displayed.
Is anything I type sent to a server?
No. Every keystroke is read and processed inside the page you're looking at, and the tester doesn't upload what you press. Your best runs and session stats live in localStorage on your own device, which means clearing your browser data clears them too. This matters more for a keyboard tester than most tools, because the whole point is testing real, sometimes password-adjacent keys — there's no reason that data should leave the tab. The site itself does count page views and measures traffic in aggregate, which the privacy page describes in full.
What's the difference between ghosting and chattering?
Ghosting is when pressing several keys together stops one of them from registering; chattering is when a single press of one key registers as two or more. They have different causes and different fixes. Ghosting comes from how the keyboard's switch matrix is wired — press three keys that share a row and column and the controller can't always tell which ones are genuinely down, so anti-ghosting or N-key rollover hardware exists specifically to prevent it. Chattering is usually a worn or dirty individual switch bouncing on contact, which is a hardware fault in one key rather than a matrix limitation. Run the ghosting test to check the first; tap one key rapidly twenty times to check the second.
A key stays pale even though I'm definitely pressing it — what does that mean?
It means the browser never received a keydown event for that switch, which almost always points at the keyboard rather than the page. Before assuming the switch itself is dead, confirm you've selected the layout that actually matches your board — a key selected under the wrong layout can appear to test the wrong physical position. If the layout is correct and the key still won't light, try reseating the cable, testing the board on another machine, or checking whether an entire row or cluster is out rather than just one key; a whole row failing together usually means a connector or ribbon problem, not one bad switch.
Which layout should I pick — 100%, 80%, 75%, 65% or 60%?
Count what your keyboard has on the right-hand side, because that is what the percentages describe. A 100% board is full-size: function row, a navigation cluster of Insert, Delete, Home, End, Page Up and Page Down, and a 17-key numpad. An 80% board — usually sold as tenkeyless or TKL — is the same thing with the numpad removed. A 75% keeps the function row but squeezes the navigation keys into a single column down the right edge, so the whole board is one solid rectangle. A 65% drops the function row and keeps the arrows and a short navigation column. A 60% keeps only the main block: no function row, no arrows, no numpad. Pick the one whose shape matches the board in front of you and the on-screen keys will line up with your keycaps.
What is the difference between ANSI, ISO and AZERTY?
They describe the shape and the printing of a board rather than its size, which is why this tester asks for them separately. ANSI is the US shape: one long left Shift and no extra key beside it. ISO, standard across the UK and Europe, takes a keycap's width off that left Shift to make room for an extra physical key, and grows the Enter key into a tall L. AZERTY is not a shape at all but a French arrangement of the printed legends, normally on an otherwise-ISO board — so a French keyboard is usually an ISO board running AZERTY. Because size and standard are independent, every combination is a real thing you can own: an ISO 75% and an ANSI 100% are both ordinary boards, and you set them with the two pickers above the keyboard.
Why is my numpad not working when I press the number keys?
Nine times out of ten the culprit is Num Lock. With Num Lock off, the numpad keys send their secondary function — Home, End, Page Up, the arrows — rather than digits, which looks exactly like a dead number pad in any application that expects numbers. Press Num Lock once and try again. Select the 100% size above and this tester will show you every numpad switch, including Num Lock itself: if the numpad keys light up on the board here but produce nothing in your spreadsheet, the switches are fine and the problem is Num Lock or a software setting. If they do not light up at all, and Num Lock does not either, the fault is in the keyboard. On a laptop with no separate numpad, some models overlay one on the letter keys behind an Fn layer, which the browser cannot see at all.
Can I test for multi-key rollover on this page?
Yes — the ghosting test built into this tool is exactly that check. Hold three keys down at once, on the same row where possible, and confirm all three stay lit at the same time; if one drops out while the other two hold, that combination is being blocked by the keyboard's matrix rather than by anything in the browser. Cheaper keyboards typically support 2- or 3-key rollover reliably and start dropping combinations beyond that, while mechanical boards marketed for gaming usually claim N-key rollover (NKRO), meaning every key should register regardless of how many others are already held.