How Accurate Is an Online Stopwatch?
A stopwatch online can look more precise than it is. Two digits after the seconds suggest a tight little machine. Those digits are a display format. They are not a certificate. Splitmark Stopwatch shows hundredths because the format HH:MM:SS.CS is easy to read for everyday sessions. It does not claim guaranteed hundredth-of-a-second accuracy, and it is not certified timing equipment.
This article explains how the browser clock is built, what happens when a tab is hidden, and how to use the tool without confusing pretty digits with laboratory measurement.
Timestamps, not a ticking counter
The elapsed-time engine stores a start timestamp from performance.now() and, while the stopwatch is running, computes elapsed time as the difference between “now” and that start, plus any time already accumulated before a pause. Display updates use requestAnimationFrame so the digits refresh while the page is visible.
That design avoids a fragile pattern: adding 10 milliseconds on a repeating timer and hoping every tick arrives on time. If a frame is delayed, the next frame still reads the real elapsed difference. You do not “lose” seconds because a few animation frames were skipped. You may, however, see the display jump slightly to catch up, which is honest. The stored elapsed time is the timestamp math, not the number of times the screen painted.
Internal math uses milliseconds. One second equals 1,000 milliseconds. The face shows hundredths, and one second equals 100 hundredths. Showing fewer digits than the computer can count is normal. It is also the first place people over-read the tool.
What “hundredths” mean on a screen
Hundredths are a rounding of the millisecond total into two decimal places. They make short laps easier to compare in a table. They do not mean the laptop crystal was calibrated this morning. They do not mean two phones will agree to 0.01 seconds after a long session. Device clocks differ. Browsers differ. Load on the machine differs.
For boiling pasta, timing a rehearsal, or seeing whether a homework section took eight minutes or eighteen, that disagreement rarely matters. For photo-finish sport, scientific trials, or any setting that requires certified measurement, it matters completely. Use the appropriate instrument there. Use this site for ordinary elapsed time in a browser tab.
Hidden tabs and sleeping devices
Browsers throttle background tabs. Animation frames may pause. Splitmark Stopwatch treats that as expected: when the page becomes visible again, it recalculates elapsed time from the timestamp clock rather than assuming every frame occurred. A long-running session that you left in a background tab should still show a total close to wall time when you return, provided the browser process kept running.
A sleeping phone or a closed laptop is a different situation. JavaScript does not keep a guaranteed schedule while the machine is asleep. A countdown that was supposed to chime may simply resume later and complete, or it may sit until the tab is awake. The completion tone, if you enabled it, plays in the open tab. It is not a system alarm. This site does not ask for notification permission in order to pretend otherwise.
If you need a timer to wake you, use the clock that the operating system provides for that job. If you need a visible remaining-time panel while you sit at the desk, the countdown timer is appropriate.
Pause, resume, and human error
Most “inaccuracy” in everyday timing is not the browser. It is a late Start, a forgotten Pause, or a lap pressed a second after the event. The engine will faithfully record the timestamps it was given. If you start the online stopwatch after you already began the task, the display will under-report. If you leave it running through a conversation, it will over-report. Pause exists so interruptions can be excluded. Reset exists so the next job does not inherit the last one.
Lap sums should match cumulative total time within display rounding. If they do not, you are probably mixing paused gaps with mental arithmetic, or reading lap time as if it were total time. The columns are defined in the laps guide.
What you can reasonably expect
You can expect a stable count-up while the tab is open and the device is awake. You can expect pause and resume to preserve accumulated time. You can expect laps to be computed from the same elapsed clock. You can expect the CSV and clipboard export to match the table. You should not expect legal-grade timestamps, synchronized multi-device timing, or a promise that hundredths are true in the physical world.
Conversions stay ordinary: 1 second = 1,000 milliseconds; 1 second = 100 hundredths; 1 minute = 60 seconds; 1 hour = 60 minutes. Mixing those units in your head is how people misread a table, not how the engine fails. The engine keeps one millisecond total and formats it for the face and the ledger.
Read the face as a well-behaved household instrument translated into a web page. That is the job. For how to operate the controls without worrying about the crystal, see How to Use an Online Stopwatch for Everyday Tasks.