It measures how many milliseconds pass between the signal and your press
The reaction time test times, in milliseconds (ms), the gap between the moment the pad turns green and the moment you press. A set is 5 or 10 trials, and the median, the middle value of the set, is shown as your headline result. There are 1,000 ms in a second, so 250 ms is 0.25 seconds.
The result is more than one number. On the reaction scale, each trial of this set is plotted as a dot next to reference values such as the athletics false-start limit, the lab average and the web median. Past sets recorded with the same input and trial count stack up as bars, so you can see whether today is faster or slower than usual.
When it is useful to measure your reaction time
Reaction time shifts with how you feel and what equipment you use. Measuring under the same conditions several times lets you see those shifts in numbers.
- When you feel sleepy or tired: sleep research reports that with too little sleep, reactions get slower overall and missed signals become more frequent. Comparing with your usual results gives a rough idea of how you are today. It is not a diagnostic test.
- After changing a mouse, monitor or wireless device: a computer and screen can add tens of ms to what you see. If the same person measures with only the equipment changed, the difference due to equipment shows up.
- Before gaming or sports: a few sets show whether your results have settled, which helps you judge whether you are warmed up.
- When competing with friends or family: after a set, send the result link. Your median appears on the recipient's reaction scale, so you can compare on the same screen.
How to measure
- Choose the number of trials (5 or 10). If it is your first time, you can press Practice run to see the signal first. Practice results are not part of the set.
- Press the pad and it turns dark red while the wait begins. The wait is random each time, between 1.5 and 6 seconds, weighted toward the shorter end so that a long wait does not tell you the signal is about to come.
- When the pad turns bright green and shows "Now!" with a filled square, press right away. Use a mouse click, a screen tap, or the space bar or Enter. The input used for the first result of a set becomes that set's input.
- Check the trial result, and when you are ready, press the pad to start the next trial. The next trial never starts by itself.
- When the chosen number of trials is complete, the set ends and the median, mean and spread appear along with your saved past sets.
To concentrate, press Full screen. The pad fills the screen, and the space bar can both start a trial and react.
How to read the result
The mean is the sum of all trials divided by their number, and the median is the middle value when the results are sorted from smallest to largest. With an even count such as 10, it is the mean of the two middle values. If your mind wanders once and you press late, the mean rises a lot while the median barely moves, so this tool uses the median as the headline result and as the basis of your best.
| Set | Trial results (ms) | Mean | Median | Spread |
|---|---|---|---|---|
| A: one late press | 200 · 240 · 260 · 280 · 520 | 300 | 260 | ±126 |
| B: steady presses | 240 · 250 · 260 · 270 · 280 | 260 | 260 | ±16 |
Both sets have the same median of 260 ms, but because of the last 520 ms, the mean of A is 40 ms higher. The spread is the standard deviation, which shows how widely the trials scatter around the mean. In Woods et al. (2015), which measured 1,469 people, the standard deviation within a person averaged 40.0 ms, so it is normal for your results to vary by tens of ms from trial to trial.
Median = middle value of the sorted results (for an even count, the mean of the two middle values)
Spread (standard deviation) = √(Σ(result − mean)² ÷ (trials − 1))
Distance covered meanwhile = speed (ft/s) × median (seconds)
Mean, median and standard deviation are calculated from the unrounded measurements and are rounded to whole ms only for display. "Distance a car at 60 mph covers meanwhile" is the distance traveled during the median. Since 60 mph is 88 ft per second, 260 ms covers about 22.9 ft.
How many ms counts as fast?
To compare reaction times, first look at what was measured and with which equipment. Values measured with lab equipment and web results measured on people's own computers can differ by tens of ms even for the same person. The reference values shown on the reaction scale are below.
| Reference | Value | What was measured |
|---|---|---|
| Athletics false start | under 0.100 s | The start information system flags a possible false start when the reaction time after the gun is shorter than 0.100 seconds (World Athletics Technical Rule 16.6). |
| College students, light | about 190 ms | The average simple reaction time of college-age people, accepted for about 120 years. For sound, it is about 160 ms (Kosinski literature review). |
| Computer study | 231 ms | The average of 1,469 people aged 18 to 65 on a calibrated computer. Subtracting the screen and mouse delay (17.8 ms) gives 213 ms (Woods et al. 2015). |
| Web median | 273 ms | The median of the results collected on Human Benchmark. It includes each person's own computer and screen delay. |
| PVT lapse | 500 ms or more | In the 10-minute psychomotor vigilance test (PVT) used in sleep research, the threshold for counting a response as a lapse of attention (Basner and Dinges 2011). |
If you measure 250 to 300 ms at home, you are near the web median. The closer to 200 ms, the faster, though it may also mean your setup adds little delay. This tool gives no age grades or "top X percent" rankings. The delay added differs from device to device, so ranking yourself against other people's results makes it easy to mistake equipment differences for differences between people.
Delays mixed into the measurement
The result on screen is your reaction time plus the time added by equipment. Human Benchmark says a computer can add 10 to 50 ms and some TVs up to 150 ms, and Woods et al. (2015) summarize that hardware and software delays can inflate measurements by up to 100 ms.
- Screen refresh rate: a screen shows a new picture only at fixed intervals. Even if the signal changes, it appears at the next refresh, so a screen with a shorter refresh interval makes you wait less before the signal is visible.
- Input device: a mouse, a keyboard and a touch screen each take a different time to report a press to the computer. That is why this tool keeps results and bests separately for each input.
- Browser time: to prevent timing attacks and device fingerprinting, a browser may reduce the precision of event timestamps. Even if the screen shows 1 ms steps, that does not mean 1 ms accuracy.
| Refresh rate | 60 Hz | 120 Hz | 144 Hz | 240 Hz |
|---|---|---|---|---|
| Refresh interval (1,000 ÷ rate) | 16.7 ms | 8.3 ms | 6.9 ms | 4.2 ms |
This tool takes the start of the frame after the one that drew the signal as the signal time, and the time the input event occurred as the press time. So even if the browser is busy and handles your press late, that waiting time does not enter the result. "Screen ○○ Hz" in the settings line is the interval at which the browser draws the screen, measured and displayed. The moment the display panel actually emits light and the delay of the input device cannot be measured, so no fixed value is subtracted as a correction.
What changes reaction time
- Practice: on an unfamiliar task, results are uneven at first and become steadier with practice. One study found that three weeks of practice shortened the reaction time to a visual stimulus, and that the effect lasted at least three more weeks (Ando et al., cited in the Kosinski literature review).
- Sleepiness and fatigue: of the forms of fatigue, sleepiness is summarized as having the largest effect on reaction time. In sleep-deprivation experiments, reactions became slower overall and lapses of 500 ms or more kept increasing.
- Age: among the 18- to 65-year-old participants in Woods et al. (2015), each additional year of age meant 0.55 ms slower on average. Between ages 20 and 60, that is a difference of about 22 ms.
- Right after a mistake: after pressing before the signal, people tend to press more cautiously on the next trial, which slows the result. If one trial after "Too early" is unusually slow, this may be why.
- Distractions: noise or doing something else at the same time lengthens reaction time. Turn off notifications and look only at the screen so that results are comparable.
What counts as too early or cancelled
- Pressing before the signal: counted as too early, and the same trial is repeated. It is left out of the set results, and only the count is shown separately.
- Under 100 ms after the signal: treated as a guess at the signal rather than a reaction to seeing it, and counted as too early. The athletics false-start rule and the PVT false-response criterion use the same 100 ms.
- No press for 10 seconds after the signal: that trial is cancelled. Trials already measured stay.
- Switching to another window or tab: if you leave the screen during the wait or the signal, that trial is cancelled, so that time spent looking elsewhere does not get into your results.
- Changing the input during a set: if a set started with the mouse and you press with the keyboard, that trial is not accepted. To change the input, press Start over.
Saved records and the result link
When you finish a set, the record is saved in this browser. For each setup, a combination of input (mouse, touch, keyboard) and trial count (5 or 10), the latest 20 sets and the best median are kept separately, and no measurements are sent to a server. If this browser holds results measured with an earlier version of the reaction time test, they are moved over once when you first open the page.
Copy result copies the median, mean, fastest, slowest and each trial as text, and Copy link copies an address that carries this set's trial results. A person who opens the link sees those results, and if they press the pad and measure their own, they can compare with the received result (◆) on the reaction scale. Clear records asks for confirmation and then deletes the records of every setup saved for this tool, without touching the data of other tools.
Limits
- It measures only simple reaction time: one visual signal and one kind of input. It cannot be compared directly with choice reaction time, where you pick among several signals, or with results for sound.
- The results include delays from the browser, the screen and the input device, and this tool neither measures nor subtracts them. A difference of a few ms from a result measured on another device is hard to read as a difference in ability.
- With only 5 to 10 trials the sample is small, so even the same person can get a different median in each set. Look at sets from several days rather than a single day.
- It is not a test of health, cognitive function or fitness to drive. The PVT in sleep research is a separate 10-minute test with signals at 2 to 10 second intervals.
- In real driving, the time to recognize a hazard, decide and move your foot is added before the brake is pressed, so it is not the same value as these results.
- Records stay only in this browser. They are not visible if you clear the site data or open the page on another device.
Reference date Sources were checked on 2026-10-21.
Sources
- 01World Athletics: Competition Rules & Technical Rules (C1.1 & C2.1)Technical Rule 16.6: the start information system flags a possible false start when the reaction time is under 0.100 seconds
- 02Basner, Dinges (2011) Maximizing sensitivity of the Psychomotor Vigilance Test (PVT) to sleep loss, SLEEP 34(5)Lapse of 500 ms or more, false response under 100 ms, 10-minute test with 2 to 10 second intervals, slower responses and more lapses with sleep loss
- 03Robert J. Kosinski: A Literature Review on Reaction Time (Clemson University, archived copy)Simple reaction time of college students: about 190 ms to light, about 160 ms to sound; effects of practice, mistakes, sleepiness and distraction
- 04Woods, Wyma, Yund, Herron, Reed (2015) Factors influencing the latency of simple reaction time, Frontiers in Human Neuroscience 9:1311,469 people aged 18 to 65, mean 231 ms (213 ms corrected for equipment delay), 0.55 ms per year of age, within-person standard deviation averaging 40.0 ms
- 05Human Benchmark: Reaction Time TestMedian of collected results 273 ms; a computer can add 10 to 50 ms and some TVs up to 150 ms
- 06MDN: Window.requestAnimationFrame()Calls usually match the screen refresh rate and mostly pause in background tabs
- 07MDN: Event.timeStampThe time the event was created; precision may be reduced to prevent timing attacks and fingerprinting
Frequently asked questions
What is the average reaction time in ms?
It depends on how it was measured. The average for college students to a light signal with lab equipment is about 190 ms, the average of 1,469 people aged 18 to 65 on a calibrated computer is 231 ms (213 ms after subtracting equipment delay), and the median of Human Benchmark web results measured on each person's own computer is 273 ms. A result you measure at home is best compared with the web median, which was measured the same way.
Why does the test highlight the median instead of the mean?
If your mind wanders once and you press late, the mean rises a lot while the median barely moves. For example, 200, 240, 260, 280 and 520 ms have a mean of 300 ms but a median of 260 ms. This tool sets your best by the median and shows the mean and spread alongside it.
Why is a press under 100 ms repeated?
It is treated as a guess at when the signal would come rather than a reaction to seeing it. In athletics, the start information system flags a reaction under 0.100 seconds after the gun as a possible false start, and the PVT in sleep research also counts under 100 ms as a false response. Such a press counts as too early and the same trial is repeated.
Does practice make reaction time faster?
To some extent it makes it faster and steadier. On an unfamiliar task, results are uneven at first and settle with practice, and one study found that three weeks of practice shortened visual reaction time. It may just be getting used to this screen, though, so it is hard to say that other things became faster too.
Why are touch results on a phone slower than on a computer?
Devices differ in how long the screen takes to show the signal and report the press, and the way a finger moves is different too. Human Benchmark says a computer can add 10 to 50 ms. That is why this tool keeps mouse, touch and keyboard results and bests separately.
Does being tired or sleepy slow my results?
It can. Sleepiness is summarized as the form of fatigue with the largest effect on reaction time, and in sleep-deprivation experiments reactions became slower overall and lapses of 500 ms or more increased. This test is short, at 5 to 10 trials, and is not a diagnostic test, so treat it only as a reference.
Does a high refresh rate monitor improve my results?
The wait before the signal appears on screen gets shorter, so results can come out slightly lower. One refresh takes 16.7 ms at 60 Hz and 6.9 ms at 144 Hz. The difference is not that size every time, though, so do not subtract a fixed value to compare; compare results from the same equipment.
Where are my records stored and how do I delete them?
When you finish a set, it is saved only in this browser and is not sent to a server. For each input and trial count, the latest 20 sets and the best median are kept, and Clear records asks for confirmation and then deletes all of this tool's records. Clearing the browser's site data also removes them.