What the test measures and what you will see on screen
Key rollover is the number of keys a keyboard can register at the same time before it starts dropping or garbling inputs. The number printed on the box is a claim; the number you measure is the truth for your keyboard, your cable or wireless link, and your computer together. This guide is a practical protocol for getting that number and, more usefully, for finding out which specific combinations fail. If you want the theory behind why keys drop, the key matrix and ghosting are explained in our keyboard rollover and ghosting guide; here we stay on the how-to.
Our keyboard test page gives you five readouts that the protocol below relies on. A virtual keyboard lights up every key the browser currently sees as held down. A currently-pressed list shows the same keys as text, which is handy when you are holding so many that the picture gets busy. A max simultaneous keys counter records the highest number of keys ever held at once during the session, and this is your rollover figure. A tested-keys count tracks how many distinct keys you have pressed, and a recent input history shows the order in which presses arrived, which is how you catch a drop that happened too quickly to notice.
The test identifies keys by their physical position code rather than by the character they type, so it works the same on US, UK, Korean, German, or any other layout, and it does not care whether you have remapped keys in software. One practical consequence: keys the operating system reserves for itself, and the Fn key on most keyboards, never reach the browser at all, so they will not light up. That is expected and not a rollover failure, and the protocol accounts for it.
Step 1 and 2: get focus, then find your ceiling
Step 1 is to click once anywhere on the test page so the browser tab has keyboard focus. If the virtual keyboard does not react to a single key press, focus is elsewhere: click the page again, make sure no other window is in front, and close any browser extension popup. Press a few single keys and confirm that each one lights and appears in the history. That confirms the basic path from keyboard to browser before you start stacking keys.
Step 2 finds your ceiling. Put one finger on a letter key and hold it. Add a second key with another finger, then a third, and keep adding while all previous keys stay down. Watch the max simultaneous keys counter: it climbs with each key that registers and stops climbing at the first key that does not. Use both hands and spread the keys across the board rather than choosing neighbours, because keys that sit together in the matrix are more likely to conflict and you want to measure the keyboard's best case first. Release everything, then repeat two or three times with different key sets. The highest number you reach is your effective rollover.
Compare the counter with what your eyes see. If you are holding eight keys but the counter says six and only six are lit, the keyboard or its link is capping at six. If you are holding eight, six are lit, and one of the lit keys is a key you are not touching, you have just seen a ghost, which we come back to in the ghosting section. Note the number and the pattern; the rest of the protocol is about turning that raw ceiling into something you can act on.
Step 3 and 4: real gaming combos, then the modifiers
Step 3 is the part that actually matters for play, because a keyboard with a high ceiling can still block a specific low-count combination. Test the combos you use. For a shooter, hold W and A or W and D for diagonal movement, add Shift for sprint, Space for jump, and Ctrl for crouch, and then tap a reload or ability key on top. For a MOBA or action game, hold Q, W, E, and R together, add Space and a number key. Run the arrow cluster with Shift and Ctrl for games that use it, and hold four or five numpad keys at once if you play anything that binds there. In each case look at the virtual keyboard and the pressed list: any key you are physically holding that is not lit is the key that drops in that combination.
Write down each failing combination and which key was lost, not just that something failed. Keyboards block by matrix position, so the losing key is often the same one every time in a given cluster, and that is the information you need later to rebind around the problem. If a combination works only when you press the keys in a certain order, that is also a rollover artefact worth noting: some controllers accept the first N keys and ignore later ones, so which key gets dropped depends on which you pressed last.
Step 4 handles the modifier keys separately, because the USB keyboard protocol does. In the standard report a keyboard sends, Ctrl, Shift, Alt, and the Windows or Command key on both sides are packed into a single bitmask that travels alongside the ordinary key slots rather than occupying them. That means on a typical six-key keyboard you can hold Shift, Ctrl, and Alt in addition to six letters, and it also means a modifier will almost never be the key that drops. Test it explicitly: hold your six-letter set, then add Shift and Ctrl and confirm they light without knocking a letter out. If adding a modifier does remove a letter, the keyboard is using a non-standard report format or is in a compatibility mode, which is worth knowing before you blame the game.
Step 5 and interpreting the number you got
Step 5 applies if your keyboard has more than one way to connect. Many boards that advertise full n-key rollover deliver it only over the USB cable and quietly fall back to a six-key limit over Bluetooth or a 2.4 GHz dongle, because the wireless HID profile or the receiver firmware uses the simpler report format. Repeat step 2 in each mode on the same keyboard and record both numbers. If wired gives you a counter that keeps climbing and wireless stops at six, you have found a real, actionable difference, and the fix for a serious session is simply to plug in.
Now read the result. A ceiling of two or three keys, especially with drops in the WASD cluster, is the signature of a basic membrane or budget keyboard with no diodes in the matrix; it is fine for typing and will frustrate some games. A ceiling of exactly six with modifiers on top is the standard USB six-key rollover and is what most gaming keyboards, and many good mechanical boards, deliver by default; it is enough for the great majority of games. A counter that keeps climbing past six as long as you can find fingers, up to the whole palm test described below, means the keyboard is running in true NKRO mode.
If you own a board sold as NKRO and the counter stops dead at six, do not assume the label lied; check the mode first. Many NKRO keyboards ship in six-key mode for compatibility with BIOS screens and older systems and switch to NKRO with a key shortcut, often Fn plus a number or a dedicated toggle listed in the manual. Some switch automatically based on the connection, as covered in step 5. Others have a firmware option in the vendor software, and a firmware update occasionally changes the default. Toggle the mode, return to the test page, click for focus, and run step 2 again; a keyboard that jumps from six to unlimited after a toggle is behaving exactly as designed.
The palm test, and ghosting versus blocking
The palm test is the quick, crude version of step 2: place your whole palm, or both palms, flat across the letter keys so that fifteen or twenty keys go down at once, and watch the counter and the pressed list. On a true NKRO keyboard every key under your hand lights and the counter shows a large number. On a six-key board the counter stops at six no matter how many keys are physically down. It is a fast way to confirm NKRO mode after toggling it, and a good in-store check. Be aware that your palm may land on the Windows or Command key and open a menu, which steals focus; if that happens, close the menu, click the page again, and keep the palm on the letter rows.
What the palm test does not tell you is anything about specific combinations. It measures the best case, a large number of keys spread across the matrix, and it says nothing about whether W, A, Shift, and Space in particular survive together. A keyboard can pass a palm test with a high count and still drop a key in a tight cluster, and a keyboard with a modest ceiling can handle your exact combos perfectly. Treat the palm test as a confirmation of mode and step 3 as the verdict on playability.
Ghosting and blocking are two different failure modes and the virtual keyboard shows them differently. Blocking, which is what modern keyboards do on purpose, means a key you are holding simply does not appear: it stays dark, it is missing from the pressed list, and the history shows no press for it. Ghosting means a key you did not touch appears lit and shows up in the pressed list as if you had pressed it, because the matrix connected a fourth intersection when you closed three that share rows and columns. On most keyboards made in the last decade you will only ever see blocking, because the controller suppresses the ambiguous state instead of reporting a phantom. If you do see a ghost, note the three keys that produced it; the phantom key is always the fourth corner of the rectangle those three form in the matrix, and it will recur every time.
What to do with the result, plus a short FAQ
If a specific combination fails, the cheapest fix is a rebind. Move the dropped action to a key far from the cluster, for example shift a skill from E to Z or crouch from Ctrl to a mouse button, then return to the test page and confirm the new combination lights fully. If wireless mode was the limiter, use the cable for games where it matters. If the counter stopped at six on an NKRO board, enable the NKRO mode as described above and retest. If the keyboard is genuinely a two or three key board and your games need more, that is the one case where new hardware is the honest answer, and there is no software that raises a matrix limit.
When you are choosing a keyboard, run this protocol before you pay. Open the test page on a phone or laptop, plug the demo unit in or pair it, and do steps 2, 3, and the palm test on the spot; it takes under two minutes and tells you more than the packaging. If the shop unit is wireless-only, remember step 5 and ask whether the wired figure is different. On unboxing day at home, run the full protocol once while the return window is open, and keep the numbers.
Does rollover affect typing? Almost never. Even fast typists rarely have more than three keys down at once, and typing problems such as doubled letters or missed single keystrokes have different causes; if a single key repeats or skips on its own, see our key chatter guide, and if a key never registers at all, our keys-not-working guide covers that. Do laptops have low rollover? Frequently yes: most non-gaming laptop keyboards are simple matrices without diodes and land at two or three keys in this test, while gaming laptops and some premium models do better, so run step 3 on a laptop before assuming a game is at fault. Does the browser limit the count? No. The browser reports every key event the operating system delivers and does not impose a maximum, so a counter that stops at six reflects the keyboard, its mode, or its connection rather than the web page. The one thing the browser cannot see is keys the OS or the keyboard's own Fn layer consume before they reach it, which is why those keys stay dark regardless of rollover.