Why Your Reaction Test Says "Too Soon" (and Why That's the Right Call)
Click before the signal on almost any reaction test and you get some version of the same message: too soon, too early, false start. It can feel like the test is being unfair, especially if your click landed only a fraction of a second before the colour changed. It isn’t being unfair. It’s applying the one rule that keeps the whole measurement honest, and it’s worth understanding why, because it explains a lot about what a reaction time number actually means.
The floor is 100 milliseconds, and it isn’t arbitrary
Every test on this site rejects a response under 100 ms as anticipation rather than reaction. That number isn’t a guess. It’s the same threshold World Athletics uses to rule a sprint start a false start: below roughly 0.100 seconds, the standard position is that a human could not have processed the signal and acted on it, so a response that fast must have started before the signal existed. The full argument, including the research that disputes it at the margins, is on the reaction time world record page. The short version is enough here: 100 ms isn’t this site’s opinion, it’s borrowed from the one place reaction time is measured seriously enough to need a rule everyone agrees on.
A stored personal best under that floor gets discarded for the same reason, rather than kept as an impressive outlier. If the number is physiologically impossible, it was never a real reaction time to begin with, and keeping it would mean quietly lying to you about your own best score.
The measurement can only ever be too slow, never too fast
Here’s the part that matters even when you’re well clear of the 100 ms floor. The clock on every test here starts the moment the signal is set, which is a line of code executing. It stops the moment your click, tap or key press is registered, which is a hardware event arriving back at the browser. Between those two moments sits everything that has to happen for you to actually see the signal: the browser has to paint the new frame, your monitor has to display it, light has to reach your eyes.
None of that is subtracted from your score. It all gets added to it. Which means a measured reaction time is always at least as slow as your true biological reaction, and usually a little slower, never the other way round. A 60 Hz monitor alone can add up to 16 ms of pure display latency before your nervous system is even involved. This is also why comparing your score against a stranger’s posted screenshot is close to meaningless unless you know what they were tested on, covered in more detail on the statistics page.
The practical upshot: if a round ever comes back faster than feels possible, the honest explanation is almost never “I reacted inhumanly fast.” It’s “I started moving before there was anything to react to.”
Two different tests, two different honest answers
Not every test on this site handles an early click the same way, and the difference is deliberate rather than an inconsistency worth fixing.
On the reaction time test and its F1 and sound variants, an early click voids the round. You get told it was too soon, and the round restarts. That’s the right behaviour for a test measuring one clean signal-to-response interval: a guess that happened to land isn’t a reaction, so it shouldn’t count as one, and discarding it is the honest move.
On the peek reaction tests built for Valorant, CS2, Apex Legends and Fortnite, an early click is scored instead, as the slowest possible round, not thrown away. That looks like a harsher penalty, and it is one, on purpose. Those tests exist to approximate holding an angle in a tactical shooter, and clicking before the target actually appears isn’t a measurement error in that context, it’s the exact mistake the whole page is about: peeking on a guess instead of on real information. Scoring it as the worst possible outcome rather than quietly discarding it is what makes the test honest about what it’s actually simulating, rather than being honest about reaction time and silent about everything else a real duel punishes.
Neither choice is more “correct” than the other in the abstract. They’re correct for what each test is actually trying to tell you.
Getting a clean read
A few things actually move your number, and none of them are about outsmarting the test:
- React to the signal, don’t guess at the rhythm. Every delay on this site is randomised for exactly this reason. If you find yourself clicking on a rough rhythm rather than waiting to see the change, your average will look better and mean less.
- Take all five rounds in one sitting, without restarting to chase a lucky one. An average of five honest attempts tells you more than a single good round that happened to survive.
- Know what you’re testing on. A 60 Hz laptop screen and a 240 Hz gaming monitor will not give you comparable numbers, for the display-latency reason above. Compare your own score against your own previous attempts on the same setup, not against someone else’s.
If you want to see exactly where a given average sits against a real reference distribution rather than a guess, the statistics page has the full percentile table. And if “too soon” keeps happening to you specifically on one of the peek reaction tests, that isn’t a bug to work around. It’s the test telling you the same thing a coach would: you’re peeking blind.