Online Stopwatch
An accurate online stopwatch with lap times, running from the system clock rather than a tick count so it stays correct in a background tab.
The display cannot show every hundredth it counts
The readout ends in hundredths, which is 100 distinct values every second. A browser repaints about 60 times a second. Those two numbers do not divide, and the smaller one wins.
| Per second | |
|---|---|
| Values the last digit can take | 100 |
| Times the screen is redrawn | 60 |
| Values actually shown | at most 60 |
| Never shown | 40 (40%) |
A frame is 16.67 ms and a hundredth is 10, so each frame advances the digit by 1.67 of a step. Watch it closely and it moves by one sometimes and by two others — exactly 40 of the 60 frames in a second take the double step, and the two sizes add to a whole second of hundredths.
The measurement underneath is fine. It is read at the moment of the frame and is accurate to well under a millisecond; it is only the display that cannot show every value, which is a different and much smaller problem.
And the hand holding it is the real limit
The last displayed digit is worth 10 milliseconds. A person is not.
| Reacting to | Takes about | In last digits | In frames |
|---|---|---|---|
| a sound | 160 ms | 16 | 10 |
| a light | 250 ms | 25 | 15 |
| a choice between two | 400 ms | 40 | 24 |
So the hundredths place on a hand-operated stopwatch is about 25 times finer than the thing operating it, and about 15 frames of screen. It is not wrong to show it — the arithmetic is exact, and two identical readings really are identical — but a difference of a few hundredths between two hand-timed runs is a fact about the hands.
One thing it gets right that is easy to get wrong
Elapsed time here is measured with the monotonic clock rather than the wall clock. The wall clock is allowed to jump, and a stopwatch built on it would record those jumps as elapsed time.
| If this happened mid-run | A 10-second run would read |
|---|---|
| a network time correction | 12 seconds |
| a daylight-saving change | 3,610 seconds |
| someone setting the date | 86,410 seconds |
Even the mildest of those — a network time correction, worth 2 seconds — is two hundred times the precision the display is claiming. The worst is 86,400 seconds.
The monotonic clock only ever moves forward at the rate of the machine, which is exactly what a stopwatch wants and exactly what a calendar does not. It also means the elapsed figure does not depend on the display keeping up: repainting stops while a tab is in the background, and the next frame reads the clock afresh rather than counting the frames it missed.
How to use
- Press start to begin timing immediately.
- Record laps without stopping the clock, as often as you need.
- Pause and resume without losing the elapsed total.
- Copy or export the lap list before you navigate away.
- Reset to clear everything and start again.
Frequently asked questions
How accurate is a browser stopwatch?
Accurate to a few milliseconds for any practical purpose, because elapsed time is measured against the system clock rather than accumulated from timer callbacks that can be delayed. What is far less precise is your own reaction time pressing start and stop, which dwarfs any inaccuracy in the browser.
Does it keep running if I switch tabs?
Yes. Browsers throttle how often a background tab redraws its display, but the elapsed time is computed from timestamps rather than counted up, so it remains correct throughout. The display simply catches up the moment you return to it.
What is the difference between a lap and a split?
A lap is the time taken for one segment; a split is the total elapsed time at that point. Recording a lap here gives you both — the segment just completed and the cumulative time — which is why race timing conventionally reports the two figures side by side.
Why does my time differ from an official result?
Because official timing uses electronic triggers rather than a human pressing a button, and because chip timing in road races measures from the moment you cross the start mat rather than from the starting gun. Hand timing is conventionally considered accurate to about a quarter of a second at best.
Will I lose my times if the page reloads?
Yes — the stopwatch holds its state in the page itself, so a reload clears it. Copy or export the laps before navigating away if they matter to you.
Does this work offline?
Once the page has loaded, yes. Everything runs in your browser, nothing is sent anywhere, and no account or installation is needed at any point.
Can I use it for interval training?
It works, though a dedicated interval timer suits that better. A stopwatch counts up and requires you to watch for each mark; an interval timer counts down and signals you, which is what you want when your attention is on the exercise rather than the screen.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.