Long-run capture
Roll keys continuously until the ring closes
-- Hz
Time 30.0 sEvents 0live
Press any key or click to start.
Report interval histogram (ms, 1 ms buckets)
Last 300 intervals (dashed line is the median)
A polling rate is a headline number; stability is what you feel. Start a capture and keep several keys moving the whole time. The page measures the gap between every report, draws the histogram and timeline, and grades the result on how tight the intervals are and how many stalls, gaps far longer than the median, slipped in. Wireless boards and overloaded USB hubs show up here even when the rate looks fine. The keyboard latency test runs entirely in your browser, so nothing you press is stored or sent anywhere.
Roll keys continuously until the ring closes
-- Hz
Time 30.0 sEvents 0live
Press any key or click to start.
Report interval histogram (ms, 1 ms buckets)
Last 300 intervals (dashed line is the median)
--
Results appear when the capture ends.
Run the keyboard polling rate test to see how often your keyboard reports to your computer and whether that keyboard polling stability holds for the whole run. A keyboard polling rate tester counts every report your browser receives, then shows a max, min and average Hz, so you can tell a steady 1000Hz gaming keyboard from one that drifts.
The polling rate test measures reports per second, so it only works while keys are actually changing state. Use a wired connection first, close heavy apps, and keep the browser tab in focus until the timer ends. Run the keyboard debounce test to see your result in seconds, right in your browser.
The real-time rate updates while you press, then the test keeps the max rate (the peak), the min rate (the minimum) and the average rate for the run. A sample 10-second run on a wired board gave these numbers:
| Reading | Sample value | What it tells you |
|---|---|---|
| Max rate | 996 Hz | The maximum reporting speed seen, close to a 1000Hz setting |
| Min rate | 458 Hz | A slow stretch, often caused by pausing between key presses |
| Average rate | 874 Hz | The typical speed across the run |
| Stability | 87.8% | Average as a share of the peak |
Stability is the average divided by the peak:
$$\text{Stability} = \frac{\text{average rate}}{\text{max rate}} \times 100 = \frac{874}{996} \times 100 \approx 87.8\%$$Keyboard polling rate, also called report rate or polling frequency, is how many times per second your keyboard sends its key state to the computer over USB. It is measured in hertz, and a higher number means shorter gaps between reports. Over the USB HID standard, the keyboard checks in on a fixed schedule rather than streaming every press the instant it happens. Run the key release test to see your result in seconds, right in your browser.
Each setting has a matching report interval, the time between two reports. The report frequency and the interval are reciprocals:
$$\text{Interval (ms)} = \frac{1000}{\text{polling rate (Hz)}}$$| Polling rate | Report interval | Typical use |
|---|---|---|
| 125Hz | 8ms | Office and older boards |
| 250Hz | 4ms | Light typing |
| 500Hz | 2ms | Mid-range boards |
| 1000Hz | 1ms | Most gaming keyboards |
| 4000Hz | 0.25ms | High-end competitive boards |
| 8000Hz | 0.125ms | Enthusiast models |
These values are nominal specs for the interval, not a promise about total latency. Latency also includes switch sensing, firmware, the operating system, the game and the display, and the polling rate improves only one stage of that chain.
A mouse streams constantly changing position data, so extra reports make the cursor smoother. A keyboard only reports a keystroke when a key changes state, so one press fires one event. That is why a keyboard needs far less polling than a mouse to feel instant.
A fast peak means little if the average sags. Polling rate stability describes how consistent the report rate stays through the run, and it is the number to read first when a board feels inconsistent.
The sample 87.8% above is moderate: the 458 Hz minimum came from a pause, not a fault, so rerun with steadier pressing before blaming the board.
Browsers batch events, and OS scheduling decides when each one is delivered, so a web test can under-read a fast board. Treat the stability analysis as an estimate: a consistent value near 1000Hz means the hardware is performing well, while a reading far below your setting is a reason to look deeper at the system.
A low polling frequency in the test usually comes from the test conditions, not a broken device. Compare what you see with your keyboard's rated data before changing anything, because a slow system or a worn cable adds delay and latency of its own.
Many office devices are locked to 125Hz in hardware, so no software setting will push a standard board past roughly that figure on any system. A gaming keyboard is the device built to run at 1000Hz and above.
Dana is deciding whether a new wireless board is worth keeping for ranked shooters. The spec sheet says 4000Hz, but the return window closes Friday, so the keyboard polling rate test comes first. The dongle is plugged into a front-panel USB hub, and the keyboard is set to 2.4GHz mode.
Dana clicks Start Test, alternates A and S for the full 10 seconds and reads the result: max rate 1,002 Hz, average rate 611 Hz. Dividing 611 by 1,002 gives a stability of 61.0%, well under the 80% line that separates moderate from unstable. An average of 611 Hz also means a typical gap of 1.64 ms between reports instead of 1 ms.
The peak proves the hardware can report at 1000Hz, so the dip points at the path, not the keyboard. Dana moves the dongle to a rear motherboard port on a short extension and reruns the same test:
| Setup | Max rate | Average rate | Stability |
|---|---|---|---|
| Front hub, 2.4GHz | 1,002 Hz | 611 Hz | 61.0% |
| Rear port, 2.4GHz | 1,004 Hz | 941 Hz | 93.7% |
Changing one input, the port, lifts stability by 32.7 points. That is moderate, just short of the 95% stable mark, and a third run with steadier presses is the next step before judging the board.
My own take: the number to watch is the average, not the peak, because every board can hit its peak for a moment. One caveat matters here. A browser test tops out near 1000Hz on most systems, so this run cannot confirm the advertised 4000Hz, and a low reading never proves the board is slow when the pressing rhythm or the port is the cause. Dana keeps the keyboard, leaves it on the rear port, and checks it again after a firmware update.
These three terms are easy to mix up. The polling rate is how often the keyboard sends data to the PC, while the other two happen inside it.
The scan rate is how often the firmware checks the key matrix for a press. A fast polling rate cannot help if the scan rate is slow, because the switch change is detected late and the USB report then repeats old data.
Debounce filters the electrical noise from a switch so one press is not read as several. It adds a deliberate delay, so a keyboard with long debounce can feel slow even at 1000Hz.
Only up to a point. Polling rate sets the maximum wait before a detected key press is reported, and on average that input delay is half the report interval, so 1ms at 1000Hz averages about 0.5ms. It does not speed up the switch or the display, and total latency still includes every other stage.
Moving from 125Hz to 1000Hz shrinks the worst-case delay from 8ms to 1ms, a 7ms gain you can feel in fast gaming. At 1ms, the polling stage stops being the bottleneck in most gaming setups. Going from 125Hz to 500Hz already removes 6ms of that. For ordinary typing, any of these settings keeps responsiveness high enough.
From 1000Hz to 8000Hz the worst-case wait falls by only 0.875ms, and 2000Hz or 4000Hz sit in between. Human reaction time is around 200 ms, so for competitive gaming and esports the gain is tiny, and a 8000Hz keyboard mostly suits players who already have an ideal system.
A wireless keyboard often polls slower than a wired one, so test the same keyboard in each mode. Wireless gaming models trade some latency for freedom from the cable, and the wireless link is usually the part that limits stability.
A wired connection normally reports at the full rated speed. A 2.4GHz dongle can match 1000Hz on good wireless models, but Bluetooth usually reports at 125Hz or lower, so a Bluetooth keyboard can feel slower even when its keys are fine. If gaming latency matters, use the dongle or the cable.
More reports use more power. On a wireless keyboard, 4000Hz or 8000Hz drains the battery faster than 1000Hz, and that battery drain shortens the time between charges. Pick the lowest power setting that still feels responsive, and retest on a low battery, since a weak wireless link can pull your average down.
Your polling rate, frame rate and monitor run on separate clocks, so they do not need a fixed ratio.
A common rule says to multiply FPS by ten, but there is no required match. A 1000Hz keyboard on a 144 Hz monitor already reports several times per frame, and more reports do not draw more frames. System performance is limited by the display long before the keyboard.
Each report triggers an interrupt. At 8000Hz that means thousands per second, which can raise CPU usage and lower FPS on an older PC or a busy computer, so the system pays for every extra report. That load can also drag down the average the test records, so close heavy apps before you check stability.
For most people, 1000Hz is the dependable baseline. Use 125Hz to 500Hz for office work and save a higher polling rate for systems that can sustain it.
Most gaming boards let you switch the report rate in the maker's software or driver, usually between 125Hz, 250Hz, 500Hz and 1000Hz. After changing the setting, run the test again to measure the new result and confirm the keyboard polling stability still holds.
Keyboard polling rate, also called report rate, is how many times per second your keyboard reports its key state to the computer, measured in Hz. At 1000Hz it reports every 1 ms; at 125Hz every 8 ms.
Click Start Test, then alternate two keys such as A and S as fast as you can for 10 seconds. Holding a key down creates no new events. Copy the max, min and average readings into this calculator for a stability score.
Polling stability shows how close your average rate stays to your peak. Here it is the average rate divided by the max rate: 95% or higher is stable, 80% to 94.9% is moderate and below 80% is unstable.
Browser limits, OS scheduling, slow or uneven pressing, a USB hub, a wireless mode or a 125Hz office keyboard can all lower the result. Press faster, plug in directly and retest in wired mode.
1000Hz is enough for most players. Moving from 1000Hz to 8000Hz cuts the worst-case wait by only 0.875 ms and needs a capable CPU, so it mainly matters for top-level competitive play.
Yes. Each report triggers an interrupt, so 4000Hz or 8000Hz raises CPU usage and drains a wireless keyboard's battery faster than 1000Hz.
Polling rate is how often the keyboard sends data to the PC over USB. Scan rate is how often the keyboard's internal matrix checks for key presses. A responsive board needs both to be high.