How to Measure Reaction Time: Methods, Accuracy, and Why Screens Lie

Measuring reaction time seems trivial: show a signal, time the response. In practice, it's one of the easiest things to measure wrong. The number you see is only as honest as the method behind it — and most casual reaction tests quietly add 10 to 50 milliseconds of hidden lag that has nothing to do with your nervous system.

This guide covers how reaction time is actually measured, which methods are accurate, where the errors creep in, and how to get a number you can trust. The methods matter because they explain why your score differs across tools — and why "test properly" is more than advice. Start with a baseline on the reaction time test and read on to understand what that number really represents.

What you're actually measuring

Reaction time is the interval between a stimulus appearing and your response registering. Every measurement method is really measuring that interval plus whatever delay the equipment adds. A good method minimizes and accounts for the added delay; a bad one bakes it into your score without telling you.

The key distinction is simple vs. choice reaction, because they produce very different numbers:

• Simple reaction: one stimulus, one response. Measures raw speed (~250 ms average). This is what most online tests use.
• Choice reaction: multiple possible stimuli, you pick the right response. Adds 100–150 ms for the decision, and is more representative of real-world reaction[1].

If you compare a simple-reaction score to a choice-reaction score, the choice number will look "worse" — but they're measuring different things. Always compare like with like.

The measurement methods, compared

| Method | How it works | Accuracy | Best for | |---|---|---|---| | Lab chronometry | Dedicated hardware, millisecond timing, controlled display | Highest | Research | | Ruler drop | Catch a falling ruler; distance converts to time | Moderate | No-equipment estimate | | Online/app test | Software times screen-to-click | Good (with caveats) | Everyday testing | | Physical light board | Hit lights as they trigger | Good | Sports/athlete training |

Each has a place. Lab chronometry is the gold standard but inaccessible. The ruler drop is charming and equipment-free but crude. Online tests are the practical everyday option — accurate enough if you understand their one big weakness, which is the screen itself.

Why screens lie: the hidden latency problem

Here's the part most people never hear. When an online test "shows" a stimulus, several delays stack up between the software's timer and your eyes:

• Display refresh. A 60 Hz monitor only updates every ~16.7 ms, so a signal can wait up to a full frame before you even see it. A 144 Hz display cuts that to ~7 ms[2].
• Input lag. The monitor's internal processing, plus USB polling on your mouse/keyboard, adds several more milliseconds.
• Browser and OS timing. Software timers aren't perfectly synced to the display, adding jitter.

Together these can inflate a screen-based reaction score by 10–50 ms compared to lab hardware — and the slower your display, the worse it gets. This is why the same person scores differently on different devices, and why online reaction numbers run a bit higher than the ~250 ms lab average. It's not you; it's the pipeline.

The reason your phone and your gaming monitor give different reaction scores isn't your reflexes changing — it's up to 50 milliseconds of display and input lag that the test can't see and doesn't subtract.

The other error source: false starts

The second way measurement goes wrong is at the human end. If you anticipate the cue and click early, a naive test counts it as a blazing-fast reaction. Any score under about 120–150 ms on a simple visual test is a false start, not a record — the human visual floor is near 150 ms, so a faster "reaction" is a guess that happened to land[3]. Good tests detect and discard these; if yours doesn't, discard them yourself.

How to measure your reaction time accurately

You can't eliminate display lag on a home setup, but you can control everything else and get a consistent, comparable number:

1. Use the same device every time. Since hardware adds a fixed offset, testing on one device makes your scores comparable to each other even if they're not lab-perfect.
2. Warm up. Take 5–10 practice trials first — cold trials run 10–30 ms slow.
3. Take multiple trials and average. At least five measured attempts; use the mean, not your best or worst. Single trials are too noisy to trust[4].
4. Discard obvious false starts. Anything under ~120 ms is anticipation — throw it out.
5. Control your state. Test rested, at a consistent time of day, without distractions.
6. Match the task to your goal. Use visual for general reaction, audio for the fastest pathway, driving for brake response, and Stroop for choice reaction and inhibition.

Do this and your numbers become a reliable, self-consistent benchmark — the right way to track improvement over time even on imperfect hardware.

Why different tests give different numbers

Once you understand the methods, the discrepancies make sense:

• Audio beats visual by 30–40 ms because sound reaches the brain faster — not because one test is "wrong."
• Choice tests read 100–150 ms slower than simple tests because they measure a harder task.
• Faster displays read lower because they add less latency.
• Your first attempt reads slow because you weren't warmed up.

None of these mean a test is broken. They mean reaction time is a specific measurement of a specific task on specific hardware — which is exactly why consistency, not chasing a single low number, is what makes the data useful. For interpreting the number once you have it, see what is a good reaction time, and for the science of the floor these methods bump against, the fastest possible human reaction time.

The bottom line

Measuring reaction time well is less about the stopwatch and more about controlling everything around it: the same device, a warm-up, five averaged trials, discarded false starts, and a rested state. Your screen adds 10–50 ms of latency you can't remove — but if you keep the method constant, your scores stay honest relative to each other, which is all you need to see real change. Test properly on the visual and audio tests, and the number finally means something.

---

References

1. Hick, W. E. (1952). On the rate of gain of information. Quarterly Journal of Experimental Psychology, 4(1), 11-26. doi.org/10.1080/17470215208416600 — The Hick-Hyman law: choice reaction time grows with the number of response alternatives.
2. Plant, R. R. (2016). A reminder on millisecond timing accuracy and potential replication failure in computer-based psychology experiments: An open letter. Behavior Research Methods, 48(1), 408-411. doi.org/10.3758/s13428-015-0577-0 — Documents how display refresh and input lag inflate software-measured reaction times.
3. Pain, M. T., & Hibbs, A. (2007). Sprint starts and the minimum auditory reaction time. Journal of Sports Sciences, 25(1), 79-86. doi.org/10.1080/02640410600718004 — Establishes the ~100 ms floor below which a response is anticipation, not reaction.
4. Woods, D. L., Wyma, J. M., Yund, E. W., Herron, T. J., & Reed, B. (2015). Factors influencing the latency of simple reaction time. Frontiers in Human Neuroscience, 9, 131. doi.org/10.3389/fnhum.2015.00131 — Shows why averaging multiple trials is needed for a stable reaction-time estimate.