The fastest reaction time ever recorded

The record that does not exist

We went looking for the primary source behind the 101-millisecond claim: a journal article, a university announcement, a record body, a name. There is none. It circulates entirely among sites that measure reaction time and link to each other, and it has been repeated long enough to sound established.

We are not saying nobody has ever reacted in 101 milliseconds. We are saying that if it happened, no accessible record of it exists, and a number without a source is not a record.

What has actually been measured

Reactions to sound are measurably faster than reactions to light, and that is where the fastest verified human numbers come from. A study using surface electromyography on sprinters found the neuromuscular component of an auditory reaction under 85 milliseconds, with muscle activity beginning under 60 milliseconds after the signal. A separate study commissioned by athletics’ governing body, on seven national-level Finnish sprinters, confirmed reactions as fast as 80 milliseconds.

The fastest reaction ever formally recognised is 50 milliseconds, set in 2019 by researchers who used electrical muscle stimulation to fire the muscle directly the moment the stimulus was detected. It is a legitimate record and it is also the exception that proves the rule: the researchers’ own framing is that of a typical 250-millisecond reaction, roughly 200 milliseconds is forming the intention and only the last 50 or so is the muscle moving.

The 0.100-second rule is a decision, not a constant

In athletics, any movement detected sooner than 0.100 seconds after the gun is a false start. That threshold entered the rule book in 1991 following a 1989 congress decision, and it was based in part on limited data from a small group of Finnish athletes.

Two separate lines of evidence say the number is wrong, and they disagree about the direction. The sport’s own commissioned research, having measured reactions as fast as 80 milliseconds, recommended lowering the criterion to 80 or 85 milliseconds. Analyses of competition data argue the opposite: a study of 425 sprinters at the 2008 Olympics found mean reaction times of 166 milliseconds for men and 189 for women, with no recorded reactions at all between 100 and 117 milliseconds.

The most careful modern treatment came out of the disqualification of a hurdler at the 2022 World Championships for a reaction of 0.099 seconds — one thousandth under the limit. Fitting a model to reaction times pooled across meets, the authors estimated that a reaction below 0.100 seconds occurs about once in every 362 starts. Below 0.090 seconds it is about once in 14,600. They suggested 0.094 seconds as a threshold with a defensible false-positive rate.

Formula 1 does not publish reaction times

You will find confident figures for how quickly drivers react to lights-out — a grid average of around 200 milliseconds, a best-ever near 150. They are attributed to published start telemetry that does not exist. Formula 1 detects jump starts with a position sensor read by transponder loops under each grid slot, and the sport does not publish a per-driver reaction metric the way athletics does.

Drag racing does publish reaction times, per run, and elite drivers regularly post 0.05 seconds. That is not a human reaction: the Christmas Tree is a fixed, predictable countdown, so the number measures anticipation. It is not comparable to a test with a randomised start, which is what this site measures.

So what is the floor?

Empirically, controlled measurements bottom out around 80 to 100 milliseconds for sound and somewhat higher for light. That is not a derived constant, and you should be suspicious of anyone who presents it as one — the arithmetic of nerve conduction speed times synapse count that gets quoted around this subject does not correspond to any published derivation.

What is well established is the shape of the chain: signal detection, intention, and then electromechanical delay in the muscle itself, which has been measured across a physiological range of roughly 6 to 69 milliseconds. Add those and you get a floor in the right neighbourhood, but the honest framing is that the floor is where the measurements stop, not a wall anyone has calculated.