Human Benchmark Reaction Time vs ReactionBench
Human Benchmark reaction time and ReactionBench both time how fast you click after a screen turns green. Human Benchmark publishes a 273 ms median and a short test description. ReactionBench publishes five rounds, a median score and its full timing method. Neither site can remove your own display and input lag.
| Question | Human Benchmark | ReactionBench |
|---|---|---|
| Test format | "When the red box turns green, click as quickly as you can." | Wait for the panel to turn green, then click or tap. |
| Rounds | Saving needs "at least 5 clicks". Total rounds not publicly stated. | 5 valid rounds. |
| Which number is the score | Not publicly stated for a single run. The site reports a site-wide median of 273 ms. | The median of the 5 rounds, with your best round and spread shown beside it. |
| Timing method | Not publicly stated. The page says the result is "affected by the latency of your computer and monitor". | The stimulus is stamped at the painted frame and the click on pointerdown, both with performance.now(). See methodology. |
| Too-early clicks | Not publicly stated. | Responses under 80 ms are voided and the round is replayed. |
| Latency notes | States 10 to 50 ms of added computer delay, up to 150 ms on some TVs, and 30 ms as a typical desktop or laptop lag. | ReactionBench's own estimate (see methodology) is roughly 20 to 50 ms of added delay. It also links a full input lag guide. |
| Desktop and mobile | States that mobile taps "tend to be a bit slower than clicks". Whether results are kept apart is not publicly stated. | Desktop and mobile results are kept apart. |
| Accounts and saving | Sign up and login are offered, and you can save scores to see your history. | No account. Best scores stay in your browser's local storage. |
How the Human Benchmark reaction time test works
Human Benchmark is a site of brain games and cognitive tests; its home page lists eight. The reaction test page gives a one-line instruction: "When the red box turns green, click as quickly as you can." You click anywhere to start, wait for the colour change, and click again.
The "About the test" text on the page adds several facts. It says "This is a simple tool to measure your reaction time" and states that the average (median) reaction time is 273 milliseconds, "according to the data collected so far". Its statistics page adds that the site has recorded over 81 million reaction time clicks, with a median of 273 ms and an average of 284 ms.
The page is open about hardware. It says the result is affected by the latency of your computer and monitor, that your computer "could be adding 10-50ms on top", and that some modern TVs add as much as 150 ms. It also says scores here are faster than its aim trainer, because you can react without moving the cursor.
Some details are not publicly stated on the pages read for this article. These include the exact number of rounds in a run, which number is your score, the range of the random delay, how the start of the signal is timestamped, and what happens to a click that comes too early. This article does not guess at them. The only round figure on the page is that you can save after "at least 5 clicks".
How ReactionBench measures reaction time
The Classic Reaction Time Test uses the same basic idea: a panel turns green and you respond. The method is written out on the methodology page. Each round holds a random delay of 1.5 to 5 seconds, so you cannot time a rhythm. The start time is recorded at the frame that is actually painted, and your response is caught on pointerdown or keydown rather than on click, so the time you take to release the button is not added.
A response under 80 ms counts as anticipation, not reaction. It is voided and the round is replayed, so it never reaches your score. After five valid rounds the score is the median, with your best round and your spread shown next to it. Desktop and mobile results are never mixed, because touch input adds delay.
There is no account and nothing is sent to a server. Your best scores live in your browser's local storage on your own device. The trade-off is that your history does not follow you to another browser or computer.
Human Benchmark alternative: why scores differ between sites
If you score 240 ms on one site and 270 ms on another, that gap does not mean your reflexes changed. Human Benchmark itself says the test is affected by the latency of your computer and monitor and that a typical desktop or laptop adds about 30 ms. Every browser test inherits that overhead, because a web page cannot see how long the monitor takes to show a frame or how long the mouse waits to report a click.
That overhead is the main reason two sites can read differently on the same person. The test's own method matters too: when the clock starts, which input event stops it, how rounds are combined and whether early clicks are counted. A site that does not publish these details cannot be compared line by line, which is why the table above marks them as not publicly stated.
Our input lag guide breaks the chain down: monitor refresh and processing, mouse polling, the browser and touchscreens. The short version is that a fast wired setup adds little, a typical laptop more, and a TV or phone the most.
| Source of delay | Figure stated | Where it is stated |
|---|---|---|
| Typical desktop or laptop lag | 30 ms | Human Benchmark statistics page |
| Computer added on top of a human reaction | 10 to 50 ms | Human Benchmark reaction time page |
| Some modern TVs | Up to 150 ms | Human Benchmark reaction time page |
| Mobile taps versus clicks | "A bit slower" (no figure given) | Human Benchmark statistics page |
| Typical browser test overhead | About 20 to 50 ms (ReactionBench's own estimate) | ReactionBench methodology page |
Sites like Human Benchmark: comparing the test catalogs
Human Benchmark's home page lists eight tests: Reaction Time, Sequence Memory, Aim Trainer, Number Memory, Verbal Memory, Chimp Test, Visual Memory and Typing. It is a general cognitive test site, and reaction time is one of eight.
ReactionBench is built around reaction time. Its catalog has 19 tests, including the classic click test, spacebar, audio, an F1 start-lights test, a falling ruler, choice and go/no-go tasks, and a few memory tests such as the Chimp Test and the Number Memory Test. If you came for memory and typing, Human Benchmark covers more of that ground. If you want several kinds of reaction task with one published method, that is what this site focuses on.
| Area | Human Benchmark | ReactionBench |
|---|---|---|
| Reaction time | Reaction Time | Several reaction tests: click, spacebar, audio, F1 lights, ruler drop, choice, go/no-go and more |
| Memory | Sequence, Number, Verbal and Visual Memory, Chimp Test | A few, including the Chimp Test and Number Memory Test |
| Typing | Typing | No typing test |
| Other | Aim Trainer (aiming) | Aim Reaction Test |
| Total tests | 8 listed on the home page | 19 |
Which one should you use, and how to compare fairly
Use either site to track yourself, as long as you keep your device, mouse and screen the same. The fixed overhead then cancels out and a change in your score is more likely to be a change in you.
Do not compare a score from one site against a score from another as if they were one scale. The methods and the hardware overhead are not the same, and some of each site's method is not publicly stated. A phone result is also not comparable with a desktop result on either site.
If you want to see how a score ranks, the average reaction time article and the score guide explain what we compare against. To try other ways to test yourself, read how to test reaction time.
A note on names
ReactionBench is an independent site. It is not affiliated with, endorsed by or connected to Human Benchmark. The name appears here only so that readers can compare the two. Everything stated about Human Benchmark comes from its own public pages, quoted above and linked below, and it may change over time.
- 01Human Benchmark. Reaction Time Test (test page and "About the test" text), viewed 20 November 2026. Human Benchmark. Reaction Time Test (test page and "About the test" text), viewed 20 November 2026.
- 02Human Benchmark. Reaction Time Statistics, viewed 20 November 2026. Human Benchmark. Reaction Time Statistics, viewed 20 November 2026.
- 03Human Benchmark. Home page (list of tests), viewed 20 November 2026. Human Benchmark. Home page (list of tests), viewed 20 November 2026.
- 04ReactionBench. Methodology: how the timing works and what it cannot control. ReactionBench. Methodology: how the timing works and what it cannot control.