Speed

Reaction Time Test

Five trials, each measuring the gap between a colour change and your response. You get the median in milliseconds, every individual trial, and a plain statement of how much of that number your own equipment is responsible for.

Reaction time · 5 trials

The panel will turn red, then green after an unpredictable delay of 1.5–4.5 seconds. Respond the moment it turns green — click, tap, or press Space.

What your equipment contributed

Your display: not measuredassuming a typical 60 Hz panel
Adds to every trial: 0–16.7 msmean 8.3 ms
Input path adds: morenot measurable from a web page

A display can only change what it shows once per refresh, so the green panel did not appear the instant it was requested — it appeared at the next frame boundary, somewhere in that window. On top of that, your mouse, touchscreen or keyboard has its own polling and transport delay, the operating system queues the event, and the browser dispatches it. Those are real milliseconds in your score and they belong to your equipment, not to you. We do not subtract an estimate for them, because a web page cannot measure them and a made-up correction would be worse than an honest, slightly slow number.

How this is measured

What starts the clock

The panel turns green inside a requestAnimationFrame callback, and the timestamp of that frame is the start of the measurement. Frames are the smallest unit in which a display can actually change what it shows, so tying onset to one means we are timing from something the screen genuinely presented — not from a setTimeout that fired between frames and left an unmeasured gap before anything appeared.

What stops it

The first pointerdown or keydown event after onset. The timestamp is taken as the very first statement in the event handler, before any state is updated or anything is redrawn, so none of our own rendering work lands inside your score. Every time on this page comes from performance.now(), a monotonic clock, rather than the wall clock — which can jump.

How the trials are combined

We report the median of your five valid trials, not the mean. Reaction times are right-skewed: one lapse of attention adds hundreds of milliseconds to a single trial and would drag a five-trial average noticeably, while barely moving the median. Every individual trial is shown next to the result so you can see the spread rather than take our word for it.

What gets discarded, and why you are told

A response before the panel turns green is a false start and is discarded. So is anything faster than 100 ms — quicker than the physical floor for seeing something and moving, so it was anticipation already in flight — and anything slower than 1500 ms, which measured your attention rather than your reaction. Discarded trials are counted and named on the results panel. Filtering trials silently is the practice this site exists to be an alternative to.

What we cannot measure

Your display's refresh rate is measured here by timing consecutive frames, which gives the quantisation error shown above the test: 60 Hz → up to 16.7 ms, 75 Hz → up to 13.3 ms, 120 Hz → up to 8.3 ms, 144 Hz → up to 6.9 ms, 240 Hz → up to 4.2 ms. The input path — your mouse or touchscreen's polling interval, its transport, the operating system's event queue, the browser's dispatch — adds more on top, and a web page has no way to observe any of it. We state that it exists and we do not invent a figure for it, because a fabricated correction would be less honest than a number that is slightly too slow.

How your result compares

There is a defensible reference range for this task, so here it is. The largest well-calibrated study of simple visual reaction time tested 1,469 people aged 18–65 and reported a mean of 231 ms (SD 27 ms) as measured — or 213 ms once the authors subtracted the 17.8 ms their own equipment contributed (11 ms of monitor, 6.8 ms of mouse). Latencies rose from 217.8 ms in the youngest group to 239.1 ms in the oldest, roughly 0.55 ms per year of age.

Compare against 231 ms, not 213 ms

231 ms is the figure that still includes their equipment, and your score still includes yours — so that is the closer comparison of the two. It is still not an exact one: their 17.8 ms was measured on one calibrated rig running purpose-built software, while yours is unmeasured and, through a browser, very likely larger. Read the comparison as a rough bearing, not a placement.

No percentile is shown, here or anywhere on this site, unless a published source supports one. A mean and a standard deviation are not a distribution — reaction times are right-skewed, so placing you on a bell curve built from those two numbers would produce a precise-looking figure with nothing behind it. A percentile computed from BlinkBench's own visitors would be worse still: that sample is self-selected, unvetted, and spread across hardware we know nothing about.

Source: Woods, 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.
Last reviewed: July 2026

Frequently asked questions

Why is my score here slower than the figures published in research?

Because you are being measured through a browser and they were not. The reference study measured its own equipment delay at 17.8 ms and subtracted it; a web page cannot measure yours, so nothing is subtracted here. On top of that, a browser adds event-dispatch delay that purpose-built test software does not. A result 20–60 ms slower than a laboratory figure is expected and does not mean you are slow — it means the two numbers were taken with different instruments.

Would a higher refresh rate monitor improve my score?

It would reduce one specific source of error rather than make you faster. A 60 Hz display can only change once every 16.7 ms, so the stimulus appears somewhere inside that window; at 144 Hz the window is 6.9 ms. On average that is worth a few milliseconds of measured time. It changes what your equipment contributes, not what you contribute, which is exactly why we show the two separately.

Does it matter whether I use a mouse, a touchscreen or the keyboard?

Yes, and we do not correct for it. Each input path has its own polling interval and transport delay — a touchscreen typically reports more slowly than a wired mouse, and wireless peripherals add more again. Scores taken on different input devices are not directly comparable, including your own scores across two devices. If you want to compare runs, keep the device the same.

Why don't you show me a percentile?

Because we cannot compute an honest one. The reference study publishes a mean and a standard deviation, not the shape of its distribution, and reaction times are right-skewed rather than normal — so treating that mean and SD as a bell curve would produce a confident-looking percentile that is arithmetic fiction. We will not compute one from our own visitors either: that sample is self-selected, unvetted and runs on wildly mixed hardware. You get the raw number and a cited range to read it against.

Related tools

Number Memory Test

The longest string of digits you can hold and recall, presented one per second to match the research the score is compared against.

Open tool →Sequence Memory Test

The longest sequence of block positions you can reproduce in order, presented at the rate the board-based research used — and reported as a screen score, not a clinical span.

Open tool →Hearing Range Test

The highest frequency you can still hear, measured on a stepped ladder with silent catch trials — and bounded hard by your headphones, your volume and your device's sample rate.

Open tool →Colour Discrimination Test

The smallest colour difference you can reliably find, reported as three separate thresholds in ΔE — lightness, green–red and blue–yellow — and never finer than your display can render.

Open tool →