Short answer

Because the variation is real and it has been measured. In a study of 1,469 adults, one person's attempts scattered by about 40 milliseconds around their own average within a single sitting. A swing of that size between runs is the normal state of affairs, not a fault.

The complaint is always the same: 210 on the first run, 280 on the second, and no idea which one to believe. The answer is neither — believe the middle of several.

How much your own attempts scatter

In that study of 1,469 adults, the spread of one person's attempts around their own average was roughly 40 milliseconds, and the ratio of that scatter to their average sat between about 17 and 22 per cent. In other words, a person averaging 240 milliseconds routinely produces attempts in the low 200s and the high 200s in the same minute.

That is why our reaction time test reports the middle value of five attempts rather than your best one. A best-of collects your luckiest click; a middle value survives one bad attempt.

What actually causes the swings

  • Attention drifting for a fraction of a second — the largest single cause on a simple task
  • Sleep and time of day; the same person is not the same at 8am and at midnight
  • Caffeine, food, how warm your hands are
  • Input device and screen — a trackpad, a wireless mouse and a wired mouse are three different measurements
  • Background processes on the machine, which add delay unevenly rather than constantly

Why a memory test swings even harder

Scores that move in whole steps are noisier than they look. On the digit span test a single distracted list can cost you a full point, because the score is a length rather than an average.

The same is true of the Schulte table, where an unlucky grid — with consecutive numbers scattered to opposite corners — costs seconds through no fault of yours.

What we do to keep the noise down

Some of the variation can be handled by the test rather than by you, and it is worth knowing what we do about it. The reaction test reports the middle of five attempts, so one bad click cannot set your score. Anything under 100 milliseconds is discarded as anticipation rather than counted as a brilliant result.

The digit span test ends only after you miss the same length twice, so a single unlucky list does not finish the run, and every attempt is stored in your browser so you can see the pattern instead of one number.

How many runs you actually need

Three to five runs on separate occasions, on the same device, give you something worth calling your number. One run tells you what happened once.

This is also why our tests keep your history in your own browser. Your first run has nothing to compare against; your tenth has nine.

At what age does reaction time get worse?

It drifts slowly from early adulthood rather than falling off a cliff. Across large samples the change is a fraction of a millisecond per year, and research attributes it to slower stimulus processing and movement preparation rather than to worse reflexes.

Put plainly: the year-to-year change is far smaller than the difference between your good and bad attempts today.

Is a reaction time of 200 milliseconds good?

Yes. It sits near the fast end of the adult range, where the typical result is around 231 milliseconds and most people fall between 190 and 280. On a browser test it is a strong result, because your screen and input device add tens of milliseconds of their own.

Is 0.27 seconds a good reaction time?

It is ordinary — 270 milliseconds sits just on the slower side of the middle of that range. If several runs cluster there, that is your current number on that device rather than a problem to solve.

What variation does not mean

A swing between runs is not evidence of anything about you or your health. These tests measure how a task went, on one device, in one moment. If something about your concentration or memory worries you, that is a conversation for a doctor rather than for a website.

Try it properly

Do a run now and another tomorrow, on the same machine, and compare them with each other rather than with a stranger's screenshot. Start with the reaction time test — it takes about forty seconds.

Where this comes from

  • Woods et al., Frontiers in Human Neuroscience, 2015 — attempt-to-attempt scatter within one person of roughly 40 milliseconds, a coefficient of variation of about 17 to 22 per cent, a typical result of 231 milliseconds, and an age drift of roughly half a millisecond per year, from 1,469 adults aged 18 to 65. Checked 5 August 2026.
  • Age-related increases in reaction time, 2022 — the finding that slower responses in older adults come from slower stimulus processing and movement preparation. Checked 5 August 2026.