You tap the moment the screen changes, the test says 287 ms, and something about that feels slow. The internet says the average is 250. A friend swears they hit 180. A forum insists anything over 200 means you should not bother with competitive games.
Most of those numbers are being compared incorrectly. Reaction time is not one measurement, and the figure a phone or browser gives you contains several things that are not your nervous system.
Three Different Numbers, All Called Reaction Time
| Type | What you are asked to do | Typical lab range | Why it differs |
|---|---|---|---|
| Simple reaction time | One signal, one response | 180 to 220 ms (visual) | Nothing to decide, only detect and move |
| Recognition reaction time | Respond to some signals, ignore others | 300 to 400 ms | Adds a go or no-go judgment |
| Choice reaction time | A different response for each signal | 400 ms and up | Grows with the number of options |
That last row is the one that quietly ruins comparisons. Choice reaction time rises roughly with the logarithm of the number of alternatives, a relationship known as Hick's law, so a test with four buttons will always look far worse than a test with one. Neither result is wrong. They are measuring different work.
Modality matters as well, because the signal has to travel a different path before your motor system can act on it.
| Signal | Typical simple reaction time |
|---|---|
| Sound | 140 to 160 ms |
| Touch | 150 to 170 ms |
| Light | 180 to 220 ms |
Vision loses because the retina does a lot of processing before anything leaves the eye. This is also why a starting pistol beats a starting light, and why a rhythm game that gives you audio feedback feels tighter than one that does not.
What Normal Actually Looks Like
Large web based reaction datasets, the kind collected by public click-when-it-turns-green tests, tend to land on a median somewhere around 270 to 290 ms for adults. That figure gets quoted constantly as the human average, but it is really the average of a person plus a mouse plus a monitor plus a browser.
Treat the table below as a rough map of consumer device results rather than physiology.
| Age | Typical median on a normal device | Notes |
|---|---|---|
| 15 to 19 | 265 to 285 ms | Close to peak already |
| 20 to 29 | 265 to 285 ms | Usually the fastest decade |
| 30 to 39 | 280 to 300 ms | Decline is small and easy to offset with practice |
| 40 to 49 | 295 to 315 ms | Choice tasks slow more than simple ones |
| 50 to 59 | 310 to 335 ms | Consistency drops before the average does |
| 60 plus | 335 ms and up | Wide spread, fitness and sleep dominate |
Three honest caveats about numbers like these. The samples are self-selected, so the people taking reflex tests skew toward people who think they will do well. The hardware varies enormously across the sample. And the spread within any age band is far wider than the gap between bands, which means a well rested, practised fifty year old regularly beats a tired, untrained twenty year old.
The age effect is real, but it is gentler than the internet suggests for simple detection and steeper for anything involving a decision. Our guide on how to improve your reaction time covers which parts of that are trainable and which are not.
Your Device Is Inside the Measurement
This is the part almost nobody accounts for, and it is the difference between a lab number and a phone number.
| Source of delay | Typical cost per trial |
|---|---|
| Waiting for the next frame on a 60 Hz display | 0 to 17 ms |
| Panel pixel response and internal processing | 5 to 20 ms |
| Touch sensor sampling and event delivery | 10 to 40 ms |
| App or browser timing and rendering | 5 to 30 ms |
Added up, a consumer device commonly contributes something in the region of 30 to 100 ms to every single trial. A 120 Hz display shaves off a useful chunk of the frame wait, which is one reason the same person can score noticeably better on a newer phone without getting any faster.
The practical consequence is simple. A score is only meaningful against other scores taken on the same device under the same conditions. Comparing your phone result to a friend's laptop result, or to a published laboratory figure, tells you about the hardware more than about either of you.
There is a second trap in the other direction. Any test with predictable timing lets you fire early and get credit for it. If a test consistently waits two seconds, you will start anticipating, and the sub-150 ms results that follow are not reactions at all. They are guesses that happened to land. A trial faster than about 150 ms on a visual test should be discarded, not celebrated. That gap between predicting and reacting is worth understanding on its own, and timing vs reaction time goes through it properly.
The Number That Matters More Than Your Average
Here is the part that surprises people. In the research world, the mean is often the least interesting output of a reaction test.
The psychomotor vigilance task, the standard tool in sleep deprivation research, cares most about lapses: trials where the response takes longer than about 500 ms. After a short night, the average creeps up modestly, but the number of lapses rises sharply. The system does not slow down evenly. It stays roughly normal and then fails intermittently.
Response variability follows the same logic. Two people with an identical 280 ms median are not equivalent if one ranges from 260 to 300 and the other from 220 to 420. The second is faster at their best and much less reliable, which is worse in almost every real situation, from driving to a game that punishes one bad input.
So when you test yourself, look at three things: the median, the spread, and your worst trial. The spread is the one that responds first to sleep, alcohol and fatigue.
What Actually Moves Your Score
| Factor | Direction | Rough size |
|---|---|---|
| Short sleep | Slower, and far more lapses | Large |
| Alcohol | Slower and much more variable | Large |
| Practice on that specific test | Faster | Moderate, and mostly specific to that task |
| Caffeine when you are tired | Slightly faster | Small to moderate |
| A short warm up before testing | Faster | Small but consistent |
| Time of day | Fastest in the late afternoon | Small |
| Aerobic fitness | Slightly faster | Small |
| General brain training apps | Little change outside the trained task | Close to zero |
Two of these are worth expanding. Sleep is the single largest controllable factor, and the damage shows up as lapses long before you feel impaired, which is exactly what makes it dangerous. Our breakdown of how much sleep you need by age is a more useful lever here than any drill.
Caffeine helps, but mostly by restoring a deficit rather than pushing you past baseline, and the timing matters because caffeine stays in your system far longer than the alert feeling lasts. Testing at 9pm after an afternoon coffee measures something confusing.
The last row deserves its bluntness. Practice on a reaction task makes you better at that task. It does not generalise well, which is the same limitation that shows up when you look at whether brain training games work.
How to Test Yourself Honestly
1. **Use one device and keep it that way.** Same phone, same screen brightness, same hand, no case getting in the way. You are building a personal baseline, not a universal one.
2. **Run at least 5 to 10 trials and take the median.** A single trial is noise. The median resists the one lucky tap and the one distracted miss.
3. **Throw out anything under 150 ms.** That is an anticipation, not a reaction, and keeping it flatters your average by a lot.
4. **Insist on randomised intervals.** If you can feel the rhythm of the test, the test is measuring your rhythm.
5. **Record your spread, not just your middle.** Best trial minus worst trial tells you more about your state than the median does.
6. **Test under stable conditions.** Same time of day, similar sleep, no testing five minutes after a sprint or two hours into a session.
Follow those and the number becomes genuinely useful, because you can watch it move with your sleep, your caffeine and your fatigue instead of arguing with a stranger's screenshot.
Where a Reflex Game Fits
A tap-to-hit game is not a calibrated instrument, and it is worth saying that clearly. It measures the same combined chain as every other consumer test: your detection, your motor delay, your touchscreen, and your display, all bundled into one figure.
What it does well is the repetition. Zync is a fast tap-to-hit reflex game built around short rounds that speed up as you progress, which gives you the two things a single reaction test cannot: a lot of trials, and rising pressure. Volume is what exposes your variability, and pressure is what shows you whether your good trials survive when the round gets quick. Both are more informative than one number on a static test page.
Two limits to keep in mind. A game like this will not give you a clinically meaningful reaction time, and improving at it is mostly improving at it, not a general upgrade to your reflexes. Use it the way you would use a stopwatch on the same running route: the absolute figure is arbitrary, the trend across weeks is not.
If you want more in this category, our roundup of the best reflex and arcade games for iPhone covers the wider genre, and how to improve hand-eye coordination deals with the aiming half of the problem rather than the timing half.
The Bottom Line
The average adult reaction time on a phone or laptop test sits somewhere around 270 to 290 ms, and a decent slice of that belongs to the device rather than to you. Laboratory simple visual reaction time is closer to 180 to 220 ms, which is why the two figures should never be compared directly.
Judge your number against your own history on one device. Watch the spread and the lapses rather than the average. And treat anything under 150 ms as what it usually is, which is a good guess arriving slightly before the signal.