Home/Methodology

A benchmark is only as good as its method.

Most reaction-time sites never tell you how they time you. We do — including the parts we can't control. Here's exactly how ReactionBench turns a click into a number, and why that number is honest.

Written byTajammal Maqbool

sub-1 ms

Timing resolution

Rounds per score

Data sent to a server

Median

Statistic we report

THE TIMING PIPELINE

From signal to number, step by step

Arm a random delay

Each round waits a random 1.5–5 second delay before the signal, so you can never time it by rhythm. Any input during this window is a false start and voids the round.

Stamp the painted frame

The instant the panel actually flips to green — synced to the display's refresh — we record a high-precision timestamp from performance.now(), a clock accurate to well under a millisecond.

Catch the input

The moment your pointerdown or keydown event fires, we take a second timestamp on the same clock. Reaction time is simply the difference between the two.

Repeat, then take the median

After five valid rounds we sort them and report the middle value as your score, alongside your best round and the spread between them.

t_signal = performance.now() // the frame the panel paints green

t_react = performance.now() // pointerdown / keydown fires

reaction = round(t_react − t_signal) // sub-millisecond clock

THE HONEST BIT

What we can't control

No browser test is lab-grade, and anyone who claims otherwise is selling something. Between the green flash and your brain, and between your finger and our code, sits hardware we can't see. It adds a roughly fixed offset — here's where it comes from.

Display latency

The gap between the browser drawing green and your monitor showing it. Higher on TVs and phones, lower on gaming monitors.

5–25 ms

Input device lag

Wireless mice and membrane keyboards report later than wired mechanical gear. Polling rate matters here.

5–20 ms

Browser event delay

The tiny queue between the OS receiving your click and JavaScript being told about it.

1–10 ms

Total, typical

Roughly fixed on a given setup — which is why testing yourself over time stays fair even if the absolute number is a touch high.

~20–50 ms

None of this is something JavaScript can see or correct for. It's why online results — on this site and on every other browser-based reaction test — read roughly 20–50 ms above what the same person would score on dedicated lab hardware. A score here is a fair, repeatable benchmark. It is not a lab reading, and we won't pretend otherwise.

WHY FIVE ROUNDS

We report the median, not your best

A single attempt is mostly luck — one anticipated click can flatter you by 60 ms. So every test runs multiple rounds and reports the median: the middle value, which one fluke can't drag around. We also show your spread, because a tight, repeatable score means more than a lucky peak. A response under 80 ms is anticipation, not reaction — it's caught, voided, and replayed, never averaged in.

Median vs. mean. One 700 ms distraction wrecks a five-round average. The median simply ignores it and reports the middle — which is far closer to how fast you actually are.

PRIVACY & TERMS

Your scores never leave your device

There is no account, no sign-up, and no server storing your results. Every best score lives in your browser's local storage, on your machine only. Clear your browser data and it's gone — we never had a copy. ReactionBench is provided free, as-is, for curiosity and practice; it is not a medical or diagnostic tool.

No account, ever

Nothing to sign up for and nothing to log in to. Just open a test and go.

Stored locally

Your best scores sit in your browser's local storage on this device only.

Nothing uploaded

Results are never transmitted to us or anyone else. There is no leaderboard server.

Yours to delete

Clear your browser data — or your stats page — and every score is gone for good.

Questions

Why is my phone score slower than my laptop?

Touchscreens and mobile displays add more latency than a wired mouse on a fast monitor, usually 20–50 ms. That overhead is real and unavoidable in a browser, so we treat desktop and mobile as separate leaderboards for you.

Can I trust these numbers against a lab study?

For absolute values, no — a browser adds hardware overhead a research rig removes. For tracking your own progress and ranking against other web-test users on similar gear, yes. We're a benchmark, not a clinical instrument.

What happens if I press too early?

The round is flagged as a false start, voided, and re-armed. It never counts toward your median, so you can't game the score by mashing.

SOURCES

Every percentile on ReactionBench — the grade table, the "faster than N% of people" line under your score — is read off published reaction-time research, not ReactionBench's own visitors: we don't yet have enough of our own data to build a fair distribution, so published literature is the honest baseline until we do. Below is the full, deduplicated citation list behind every curve. If a source has a public link, its title links to it.

01Der, G., & Deary, I. J. (2006). Age and sex differences in reaction time in adulthood: results from the United Kingdom Health and Lifestyle Survey. Psychology and Aging, 21(1), 62-73. Link
02Kosinski, R. J. (2008). A literature review on reaction time. Clemson University. Link
03Woods, D. L., Wyma, J. M., Yund, E. W., Herron, T. J., & Reed, B. (2015). Factors influencing the latency of simple reaction time. Frontiers in Human Neuroscience, 9, 131. Link
04Jain, A., Bansal, R., Kumar, A., & Singh, K. D. (2015). A comparative study of visual and auditory reaction times on the basis of gender and physical activity levels of medical first year students. International Journal of Applied and Basic Medical Research, 5(2), 124-127. Link
05Hick, W. E. (1952). On the rate of gain of information. Quarterly Journal of Experimental Psychology, 4(1), 11-26. Link