Two definitions, two answers
Distance rate — pixels per second. How far the page moved. This is the everyday meaning of scrolling fast, and it is the figure most affected by your operating system: the same physical wheel motion covers three times the distance if you triple the lines-per-notch setting.
Input rate — scrolls per second (SPS). How many wheel events the browser received. This is much closer to a measure of your hand and your hardware, and it is the number the wider click-speed category compares. Its own catch is that a high-resolution wheel emits several events per notch, so it flatters some mice.
Neither is wrong. They answer different questions, which is why every run on this site reports both.
Measure your own — five seconds
Hover the box and scroll the way you normally would, not as fast as you can. The result gives you both rates plus a pixels per wheel event figure, which is the only reliable way to learn what one notch is worth on your specific setup.
Distance is the real travel of the box above, measured in CSS pixels, so it never depends on guessing what one wheel notch is worth. Scrolls per second counts browser wheel events — one physical notch can emit more than one. Your operating system’s scroll-lines setting changes pixels per second, so compare runs on the same machine.
Distance tiers — pixels per second
A 1,000 px/s run moves roughly one laptop screen height per second. These are the bands the scroll speed test rates against.
| Tier | Range | What it means |
|---|---|---|
| Slow | 0–599 px/s | Under about half a laptop screen height per second. |
| Normal | 600–1,499 px/s | Roughly one to two screens a second — ordinary browsing. |
| Fast | 1,500–3,499 px/s | Skimming pace. Several screens a second. |
| Very fast | 3,500–7,499 px/s | Hunting for the bottom of a long page. |
| Extreme | 7,500+ px/s | Free-spin wheel or a high-momentum touchpad flick. |
Input tiers — scrolls per second
These follow the convention used across the click-speed category, so a score here lines up with the number you would get from other SPS tools. The scrolls per second test rates against them directly.
| Tier | Range | What it means |
|---|---|---|
| Slow | 0–3 scrolls/s | Reading pace. Normal for careful scrolling through text. |
| Normal | 4–7 scrolls/s | Everyday browsing speed for a detented mouse wheel. |
| Fast | 8–12 scrolls/s | Deliberately quick flicks, or a touchpad with momentum. |
| Very fast | 13–20 scrolls/s | Sustained rapid input. Common on free-spin wheels. |
| Extreme | 21+ scrolls/s | Free-spin flywheel territory, or a very high wheel resolution. |
Why the pixel figure travels so badly
Between your wheel and the page there are at least three multipliers, and every one of them is configurable or platform-specific.
- The operating system
- Windows reports a fixed notch unit to applications and separately holds a “lines to scroll at a time” setting, three by default, adjustable up to a full screen. macOS has no line setting at all and instead applies an acceleration curve, so a fast wheel covers more ground per notch than a slow one. That alone makes a Mac and a PC non-comparable on distance.
- The browser
- Engines turn “three lines” into pixels using their own line-height maths, and they do not all agree. They also differ in what they put in
WheelEvent.deltaMode: some report pixels, some report lines for the same physical action. - The device
- A high-resolution wheel splits one notch into several smaller events. A free-spin wheel removes the notches entirely. A trackpad has none to begin with and adds a momentum tail after your fingers lift.
What we deliberately do not publish
You will find sites quoting a single authoritative “average scroll speed” to two decimal places. We do not, because we have not measured a representative sample and we are not going to invent one. The tier boundaries on this page are derived from the documented defaults above and from what the browser can actually observe — they are useful reference bands, not survey results.
What you can trust completely is your own measurement: the distance figures come from a real scroll container, so they are exact for your machine. The methodology page sets out the same boundary for the diagnostic side of the site.
Average scroll speed questions
Ordinary browsing sits at roughly 600–1,500 CSS pixels per second — about one to two laptop screen heights every second. Measured as input rate instead of distance, that same browsing pace is about 4–7 wheel events per second. Both figures are averages over a run; the peak second of a fast flick is several times higher.
Because distance per notch is a setting, not a property of the mouse. Windows scrolls three lines per wheel notch by default and lets you set anything from one line to a whole screen. macOS applies an acceleration curve, so a fast wheel travels further per notch than a slow one. Browsers then apply their own line-height maths on top. Change any of those and the pixels-per-second figure moves without your hand doing anything different.
Scrolls per second. It counts wheel events, so it is far less sensitive to operating-system scroll settings than a pixel distance. It has its own caveat — a high-resolution wheel emits several events per notch — but as long as you are both on ordinary detented mice it is the fairer contest.
There is no fixed ceiling, because raising the lines-per-notch setting multiplies distance without limit. For input rate the practical ceiling is the hardware: a released free-spin wheel can sustain 40–60 events per second for a second or two, and a trackpad momentum tail can exceed 100. On a detented wheel, the detent spacing itself caps you somewhere in the teens.
On a typical desktop setup, roughly 100 CSS pixels per wheel notch — but measure rather than assume. Run the test on this page and read the “pixels per wheel event” figure: it divides the real distance travelled by the number of events received, which is the only way to get the answer for your specific machine, browser and settings.
For reading, no — comprehension falls off long before your wheel does. A high scroll speed matters when you are hunting through long logs, spreadsheets or code, and it is mostly a settings and hardware question rather than a skill. If you want it faster, raise lines-per-notch before you train your finger.