Keyboard test

How to test your keyboard

This online keyboard tester shows the key events your browser receives. Use it to check a suspect key, work through a keyboard after cleaning, or inspect a second-hand keyboard before deciding what to investigate next.

1. Choose a reference. Choose your system, find a country, language or layout name, then select a reference and a format with or without a number pad. The compact option is one schematic arrangement; compact keyboards vary. Under “Adjust which keys are present”, exclude positions your keyboard does not have.
2. Activate the test. Choose Start testing or focus the diagram. Press your physical keys one at a time, working through letters, numbers, modifiers, navigation keys and the number pad. Holding a key does not increase the number of detected positions.
3. Review the observations. Find your report just below the test. Detected means the browser received a response. Not observed means more checking may be needed. Excluded means you removed that position from the selection.

Tab and Esc are observed if received, then leave the test. Browser and system command shortcuts remain available; a command such as F5 may reload the page and clear its observations. Switch back to Start testing after a shortcut takes you elsewhere. Clicking or tapping the drawing does not count as a keyboard test.

A repeatable check for a suspect key

Choose “Check one key step by step”, then select the physical position. Press and release once, make five separate taps, then hold for at least two seconds and release. Only received keyboard events advance the steps. Selecting a key on screen never validates it.

A held key can produce many native repeat events: that is normal system behaviour. We show those separately from new press/release cycles. For unwanted extra presses, compare the event timestamps between successive new presses after a release. The adjustable 40, 80 or 120 ms threshold is exploratory; fast intentional taps can also cross it. A close pair alone is not proof of a damaged switch.

Leaving the test while holding a key records an interruption and discards that unfinished hold. An unpaired new press or unusable timestamp is reported separately. Durations use the browser’s event timestamps, so a busy page does not turn its own handler delay into a false short interval. Browser input delivery, timestamp precision, the operating system and firmware still limit the observation.

Choose “Compare a key combination” to establish that each selected key responds alone, then try them together in different orders. Choose a preset or build a custom selection of two to six positions. The report gives the highest number of selected keys actually observed together and the completed attempts. A successful selection does not prove unlimited rollover or every possible combination. These checks stay on this page until reset or navigation; changing a target restarts its guided sequence, while its cycle aggregates remain until “Restart this key” or a full reset.

A key does not respond: what to check next

First check that the test says “Listening” and the position exists on your keyboard. Press the key alone. A missing highlight can mean an absent key, an application intercepting the input, a reserved shortcut, a different mapping or a connection problem.

Try the key in another application where it has a purpose. Check the keyboard language selected in your operating system. For an external keyboard, check its cable or wireless connection and battery, then compare another port or device if available. Keep the same key and action in each comparison.

A response in another application points toward a context or mapping difference. A repeated failure across applications and devices deserves further hardware investigation; this page cannot establish the electrical cause or guarantee that replacing a switch will fix it.

Why the letter can differ from the drawing

The drawing follows physical position codes. “Last key received” shows the value reported for that press, which can change with your language, Shift, AltGr or remapping. For example, the position named KeyQ commonly produces A on French AZERTY. That difference alone is not a faulty key.

The catalogue includes international Windows, Mac and ChromeOS references: regional variants, non-Latin scripts, Dvorak, Colemak and other layouts. Choosing Mac changes the layout’s characters, Option layers and modifier keys. It does not change your computer’s input settings.

Physical keys are a separate choice: ANSI, ISO, JIS or ABNT2. The initial suggestion does not identify your hardware. The diagram remains schematic; adjust the positions actually present. Some browsers swap the two special Mac ISO position codes: the dedicated option exchanges them explicitly, without guessing a layout from received text.

Windows and ChromeOS references come from the Unicode CLDR 43 archives; Mac character maps were extracted from Apple’s built-in input sources on macOS 26.6.2. The exact source is linked below the selector and included in the report. These versioned tables do not certify every keyboard model or OS version. Shift and Option/AltGr layers are references: IME composition, dead-key sequences and Caps Lock are not simulated. Identify Apple keyboard variants.

Testing the number pad and special keys

The full format separates number-pad positions from the number row. Try both Num Lock states if your keyboard supports them: the same number-pad position can send a digit or a navigation value. The displayed count uses the reported position, so Numpad1 and Digit1 remain separate when your browser exposes them separately.

Fn often acts inside the keyboard or device and may send no independent event. Media controls, power buttons, some function keys and operating-system shortcuts can also be intercepted. A key outside the selected diagram appears in the last-key readout but does not increase its coverage. If no usable physical code arrives, the tester says so instead of guessing a position.

A tablet or phone needs a connected physical keyboard for this test. An on-screen keyboard may supply text without reliable physical position codes. The optional combination mode compares selected keys alone and together. The guided key mode separates normal held-key repeats from new press/release cycles. Neither measures hardware latency nor establishes an electrical fault.

What stays on your device

Key observations stay in this page’s memory and disappear when you reload or leave. No text history is kept. The report appears below the test as observations become available. Choose “Download PDF” to create a file on your device, without a print dialog or an external conversion service. It contains the chosen reference, position lists, coverage, completed cycle counts, exploratory interval aggregates, guided observations and method version. No event timeline is retained. It excludes received characters, device identifiers and serial numbers.

TinyFingers uses its existing Google Analytics for site usage and broad actions such as beginning a test or viewing a report. Key values, position lists and typed text are never attached to those events. Read about TinyFingers and privacy.

Method: keyboard observations v2. How TinyFingers tests inputs. Physical position codes and received key values are documented by MDN. See also native key repeats and event timestamps.

Check another input

• Wrong characters and keyboard layout
• Ghosting and rollover: compare a combination
• Follow the guide for a key that does not respond
• Test mouse buttons, dragging and scrolling
• Investigate an unwanted double-click
• Explore TinyFingers Tools

Looking for playful keyboard exploration? Open the TinyFingers keyboard toy.