What Is Keyboard Polling Rate and How Does It Affect Input Latency?

Polling rate is how many times per second your keyboard reports its key states to the computer, measured in Hz. A 1000Hz keyboard sends a report every 1ms; a 125Hz keyboard sends one every 8ms. Higher polling rate reduces the maximum delay between a keypress and the computer receiving it, but it does not eliminate input latency on its own — the total delay also includes switch debounce, firmware processing, USB transfer, OS handling, and display refresh. If you are troubleshooting lag, test polling rate alongside latency, rollover, and chatter rather than treating it as the single cause.

Polling rate vs. input latency: what actually changes

Polling rate sets the interval at which reports are sent. Latency is the total time from physical actuation to the effect appearing on screen.

The key word is worst-case. If a keypress lands just after a report is sent, it waits nearly a full interval before the next one. At 1000Hz that wait is at most 1ms; at 125Hz it can be 8ms. That 7ms difference is real but small compared with other links in the chain — a 60Hz monitor alone adds up to ~16.7ms between frames, and switch debounce commonly adds several milliseconds.

So a higher polling rate shrinks one component of latency. It does not make a slow keyboard fast if the bottleneck is elsewhere.

Polling rate is not the same as NKRO or ghosting

These three terms get mixed up constantly, and they describe different things:

• Polling rate — how often the keyboard reports to the host. A timing/spec number in Hz.
• NKRO (N-Key Rollover) — how many keys the keyboard can register simultaneously without dropping any. A 6KRO board may fail when you hold many keys at once; an NKRO board registers them all.
• Ghosting — phantom or missing key registrations caused by the keyboard's matrix wiring when multiple keys are pressed together. It is a correctness problem, not a timing one.

A keyboard can have 1000Hz polling and still ghost. It can have full NKRO and only 125Hz polling. When you test, check them separately: press keys one at a time for dead switches, press combinations (e.g. WASD + Space) for rollover and ghosting, and use a latency or polling measurement for timing.

How to check polling rate and latency with an online tester

KeyboardTest.io runs in the browser with no drivers or software, and states that all tests are performed locally — key content is not recorded or uploaded. Its suite covers rollover (NKRO), input latency, polling rate, and chatter detection.

A practical sequence:

1. Basic function test — press each physical key and confirm the matching on-screen key lights up. This catches dead switches and unresponsive keys.
2. Rollover detection — hold several keys at once (e.g. WASD + Space). If some keys don't light up, the board may have ghosting or a rollover limit.
3. Latency / polling measurement — use the latency test to evaluate input response speed. The tool measures system-to-browser input latency from event timestamps, which is the practical number that matters for gaming.
4. Chatter test — if a single press registers twice, the switch may be aging. The chatter detection flags abnormal secondary triggers.

Expected result: a healthy keyboard lights every key, registers all simultaneous presses you make, and shows stable timing without double-triggers. If a key fails to light, that points to hardware; if timing is erratic, that points to polling, firmware, or a busy system.

Does a higher polling rate actually help gaming?

Sometimes, and less than marketing suggests.

• Where it helps: competitive play where every millisecond of input delay compounds with reaction time, and on high-refresh displays (144Hz+) where frame times are short enough that a 1ms vs. 8ms reporting gap is a meaningful fraction of a frame.
• Where it doesn't: if your monitor is 60Hz, your total latency is dominated by frame pacing, and going from 1000Hz to 8000Hz changes almost nothing perceptible. CPU load also rises with very high polling rates, which can increase latency if the system is already busy.
• Diminishing returns: the jump from 125Hz to 1000Hz is the big one (8ms → 1ms worst case). Beyond 1000Hz the gains shrink fast and depend on the rest of your chain.

A reasonable rule: 1000Hz is plenty for most users. Consider higher only if you have a high-refresh monitor, a fast CPU, and you're chasing the last fraction of a millisecond in competitive titles. If your current board already polls at 1000Hz, upgrading for polling rate alone is rarely worth it — check latency, chatter, and rollover first, since those often cause more noticeable problems.