Keyboard Tester

Key Rollover Test: How Many Keys Register at Once?

The tool opens in Rollover mode. Hold down a growing pile of keys and watch two numbers: how many are held right now, and the maximum that ever registered together. That maximum is your real, end-to-end rollover — keyboard, connection, OS and browser combined.

Held now: 0Max simultaneous: 0Hold down as many keys as you can at once.

Tab, Space, the arrows and the F-keys are captured here so you can test them. Keyboard-only users: press Tab twice in a row — the second press goes through and moves focus to the controls and links as usual.

Currently held keys

Nothing held. Press and hold a growing combination of normal keys (e.g. W + A + S + D, then add Space and more letters) and watch how many register at once. The max counter keeps your best result.

The count is what your keyboard, OS and browser deliver together — a low number can come from any link in that chain. Modifier keys (Shift, Ctrl, Alt) usually register beyond the 6-key limit of basic USB keyboards.

Rollover classes at a glance

Typical behaviour by class. Marketing names vary; what matters is the number this page measures.
ClassSimultaneous normal keysModifiers extra?Typically found on
2KRO2 guaranteed; a 3rd key works only in some combinationsYesBudget membrane boards, many older laptops
6KRO6Yes (separate bitmask)Standard USB keyboards; "BIOS mode" of gaming boards
NKROEvery key independentlyIncludedMechanical/gaming boards, mostly in wired mode

Two caveats worth knowing. First, a board can be better than 2KRO in most areas and still block specific three-key clusters — the limit is per matrix region, not global. Second, anything between your keyboard and the browser can lower the score: KVM switches, some USB hubs, remote-desktop sessions and Bluetooth power-saving modes are the usual suspects.

Ghosting vs blocking, in one paragraph

Inside a cheap keyboard, keys sit on a grid of row and column wires. Press three keys that form three corners of a rectangle on that grid and the controller cannot tell whether the fourth corner is also pressed — a phantom, "ghost" key. Because typing a letter you never pressed is the worst possible failure, controllers block instead: they simply ignore the third key of a dangerous trio. So what users experience as "ghosting" in games is nearly always blocking — a missing keystroke, not an extra one. Boards with a diode behind every switch (true NKRO designs) have no ambiguous rectangles and never need to block.

Worked example: proving a 6KRO limit

Hold W + A + S + D — the counter reads 4. Add Space: 5. Add K: 6. Now press J while everything stays down: the held count stays 6 and J never lights. Release K, and J suddenly registers. That clean "one in, one out" behaviour at exactly six keys is the USB boot-protocol limit — working as designed, not a defect. Repeat the run holding Shift and Ctrl too: the counter reaches 8 (6 letters + 2 modifiers), confirming modifiers ride in their own lane.

Worked example: catching real blocking

On one budget membrane board, holding Q + W registers 2, but adding E shows nothing — count stuck at 2 — while the combination Q + P + M happily reaches 3. Same board, same three fingers, different result: that is a matrix cluster limit. The practical consequence for a gamer: if your crouch key shares a cluster with your movement keys, remapping crouch one key sideways can fix "my keyboard eats inputs" for free, no new hardware needed.

How to run a fair test

  1. Use the wired connection and a direct USB port for the baseline (no hub, no KVM).
  2. Build combos from real usage: your movement cluster + jump + ability keys, or both players' zones on a shared board.
  3. Try clustered keys (WASD region) and spread keys (opposite board corners) — matrix limits show up in clusters.
  4. If the number looks low, check for a mode toggle (many boards ship in 6KRO for BIOS compatibility) and retest.
  5. Retest over Bluetooth if you use it — expect the wireless number to be the one you live with.

Missing keystrokes that appear only when typing fast — rather than when holding keys — are usually not a rollover problem at all. Check for double or dropped registrations with the keyboard chatter test, and run the full keyboard tester to make sure every individual key is alive. Laptop matrices carry their own quirks, covered in the laptop keyboard test.

Frequently asked questions

What counts as "good" rollover?

For typing, 6-key rollover plus modifiers is more than enough — you will never hold six letters mid-word. For rhythm games, competitive shooters with complex binds, or two players sharing one keyboard, higher limits matter and NKRO is the safe choice. The honest test is not the number itself but whether YOUR real combos register, which is exactly what this page measures.

My gaming keyboard claims NKRO but the counter stops around 6. Why?

Three common reasons: the board is in its compatibility (6KRO/BIOS) mode — many keyboards toggle modes with a shortcut like Fn+N; it is connected through a KVM, dock or older hub that only forwards the boot-protocol report; or it is in a Bluetooth mode with a lower limit than its wired mode. Check the manual for the mode toggle and retest wired.

Do Shift, Ctrl and Alt count toward the rollover limit?

Usually not. In the standard USB report, the eight modifiers (left/right Shift, Ctrl, Alt, Win) live in a separate bitmask, so you can hold all eight on top of six normal keys. That is why the tester can show counts above 6 on a plain 6KRO board when your combo includes modifiers.

Is ghosting the same as low rollover?

Related but distinct. Rollover is a designed limit on how many keys report at once. Ghosting is a matrix artifact where pressing three keys that share rows/columns would make the controller see a fourth, phantom key — so keyboards "block" (ignore) the third key instead. Low-cost matrices without diodes must block; a proper NKRO design has a diode per key and never needs to.