Reaction Time Test — Free Online Reaction Speed Check

Each of the 5 trials begins with a random wait of 1.5 to 3.5 seconds, so you cannot learn a rhythm and press early. The clock starts when the program triggers the green signal and stops when you touch the screen, so the time your browser and screen take to draw the green flash is part of the record. After each trial its time is shown, and you tap again to start the next, so the pause between trials is up to you. If you switch to another app or tab mid-trial, that trial is dropped, and when you resume only that trial restarts with a fresh random wait.

There is a reason the wait is random. The interval before the signal is called the foreperiod, and as Niemi and Näätänen (1981) reviewed, its length by itself changes reaction time. With a wait that never changed, you could simply press to the rhythm without watching for the signal. When the wait varies at random within a range, the longer you have already waited, the more your readiness builds that the signal is due, and responses tend to get faster. So it is natural for trials with short and long waits to differ a little, and the average of 5 trials is more trustworthy than any single time.

The big number on the result screen is the average of the 5 trials, in ms. The detailed view adds your best time (the fastest trial), a rating of the average and a bar per trial. The rating reads 'Very fast' at 220 ms or less, 'Fast' up to 270 ms, 'Average' up to 330 ms, 'Relaxed' up to 400 ms and 'Leisurely' beyond; these are Soundary's reference bands, not population norms. Your best time is the most flattering number but also the one most shaped by luck. If one bar is much longer than the rest, a brief dip in attention may have pulled your average up.

The PVT, the standard simple reaction task in sleep research, presents signals at random 2–10 s intervals for 10 minutes and counts responses under 100 ms as false starts and those of 500 ms or more as lapses (Basner & Dinges, 2011). In that study, lapses and the slowest responses separated sleep-deprived from rested people better than the mean or median did. This test does not filter such responses, so treat a trial under 100 ms as a lucky anticipation and retake, and note any trial of 500 ms or more as a sign about your state that day. With only 5 trials, though, it cannot replace a 10-minute PVT.

Reaction times measured in a browser include device delay. Pronk et al. (2020) tested web apps by detecting the screen change with a light sensor and pressing the screen or a key at set moments with a mechanical actuator. On the Android, iOS and Windows devices tested, recorded times were on average about 58–70 ms longer than the real ones, and longer still on some Mac browsers. Within most device–browser pairs, though, this extra delay was fairly constant. So differences between your own records in the same setup mean more than the absolute number, and switching phone, browser or wireless mouse can change your score even if you haven't changed.

Age shows up later than many expect. Der and Deary (2006) reanalysed data from 7,130 UK adults: simple reaction time (one signal, one response) changed little until about age 50, while choice reaction time, picking one of four responses, slowed across the adult range. So before 50, if this test gives you a slower time than usual, it is more reasonable to think first of sleep, fatigue or your device that day than of age. The same study also found sex differences, the most notable being in how variable choice reaction times were.