How long should a pomodoro be?

Twenty-five minutes, and the confident alternatives are softer than they sound. One of them was abandoned by the company that invented it.

The short answer

Twenty-five minutes of work, then five off, four times, then a break of twenty to thirty. Start there and change it only when you can say what is wrong with it.

If twenty-five minutes ends before you are properly into the work, go to 50/10. If you keep failing to start at all, go the other way, to 15/3. Those two complaints are the only ones that justify moving, and they point in opposite directions, which is why there is no better default than the one in the book.

What the original technique actually specifies

Three of the four numbers are ranges rather than the fixed values every app ships. The fourth is not a number at all, and it is the one that matters most.

The long break is the interesting row. Every timer counted below opens on 10 or 15 minutes, which is five to ten minutes short of the bottom of the range the technique gives. It has been copied from app to app for fifteen years and nobody appears to have questioned it.

The rule every timer drops

A pomodoro is indivisible. If something interrupts you and cannot be postponed, the interval is void and you begin a new one rather than resuming the old one.

No mainstream timer enforces this, including ours. It is the rule that makes the count worth keeping. Four protected blocks and four blocks with three interruptions in them are not the same day, and a pause button lets you record them as though they were. If you take one thing from the original technique and ignore the timings entirely, take this one.

Cirillo's own timer is the exception, and it shows how differently the technique behaves when the rule is real. It will not start until you have named the activity, there is no pause, and the button that abandons an interval is labelled Squash.

Where 50/10, 52/17 and 90/20 came from

All three are presented as upgrades. Only one of them has any data behind it, and that data no longer says what the rule says.

50/10 has no source

Nobody claims one. It is 25/5 doubled, and it survives because doubling is easy and because an hour-long unit fits a calendar. That is not a criticism: it is the alternative most people settle on, and a schedule that fits your diary is one you will actually keep.

52/17 was a company blog post, and the company has since revised it

In 2014 DeskTime, which sells time-tracking software, looked at the top ten per cent of its own users by productive-app usage. Those users averaged 52 minutes of work to 17 minutes of break. That is the whole provenance. It describes what a self-selected group of tracked office workers did on a given day. It was not a trial, there was no control, and nothing was varied on purpose.

The part almost nobody repeats is what happened next. DeskTime ran the same analysis again and published the results in October 2024:

The company that coined 52/17 measured 112/26 seven years later, and says so in public. A ratio that moves by that much when you re-run it is measuring the habits of a workforce, not a property of human attention. Use it if you like the shape of it. Do not use it because you think it is the answer.

90/20 is a sleep cycle, borrowed

The ninety minutes come from Nathaniel Kleitman's basic rest-activity cycle, a pattern first described in sleep research and later argued to continue through waking hours. The step from that to "work in 90 minute blocks" is an extrapolation made by productivity writers rather than a finding.

It is still worth trying if your work is the kind that takes half an hour to get into. Ninety minutes is also long enough that a single interruption ruins the block, which is the practical reason most people who try it go back to fifty.

What eight timers actually default to

Every one of these was opened and read on 30 August 2026, in Chrome 151 on macOS 26.6, with no account and no settings carried over. Where a timer offers presets, the value recorded is the one it opens on.

Eight out of eight open on 25 minutes, and seven of the eight pair it with a 5 minute break. Every long break is 10 or 15 minutes, which is below the 20 to 30 the technique specifies, across tools that share no code. The field agreed on one answer a long time ago and stopped asking the question.

Two rows are worth a second look. pomodoro-tracker.com is the only one that ships named alternatives, offering 30/2 and 50/10 beside the classic. Cirillo's own timer is the only one that will not let you change anything, which is the strongest signal available that 25 is meant as an instruction rather than a suggestion.

Does the timer keep time while the tab is hidden?

Choosing an interval is pointless if the thing measuring it loses time while you work. Every browser throttles a hidden tab's JavaScript timers to save battery, so a countdown built by subtracting a second per tick should undercount exactly the period you were concentrating. That was the expected result. It is not what happened.

Method: each timer was started on a fresh 25 minute focus block and hidden behind another tab from that moment. Each display was then read while still hidden, 14 minutes and 24 seconds later. Chrome 151, macOS 26.6, 30 August 2026.

All six were right. If you have read that browser timers cannot be trusted in a background tab, that advice is out of date for these six. The honest thing is to say so rather than sell a fix for a problem they do not have.

One result did not make the table. In an earlier run the same day, tomatotimers.com read 44 seconds slow after 25 minutes hidden, and 12 seconds slow after 10. It was not reproducible in the run above under conditions we could control, so it is reported here and left out of the numbers.

Does it survive closing the tab? Mostly not

This is where they come apart. Same six timers, each about fifteen minutes into a 25 minute block, each page reloaded once.

The two that survived write the finishing instant to the browser's local storage and read it back on load, so the reload costs nothing. The gap between "09:06" and "09:38" is the thirty seconds the reload itself took, which is the point: the clock kept running while the page did not exist.

Four of the six threw away sixteen minutes of work you had already done. That is a far more common way to lose a block than throttling, because an accidental refresh, a crashed tab and a browser restart all do it.

What the research tested, and what it did not

There is evidence for the method. There is none for the number, and the difference matters when someone tells you 52 minutes is scientific.

So the defensible claim is narrow and still useful: deciding the length in advance beats deciding it in the moment, because in the moment you will always work too long. Which length you decide on is up to you.

Shorter blocks, and ADHD

The advice you will find is to shorten the interval, and the reasoning is worth understanding because it is not about how long anyone can concentrate.

The expensive part of a work block is starting it. A 15 minute commitment is cheaper to agree to than a 25 minute one, so more of them actually begin, and a block that begins beats a better-shaped block that does not. The externalised timer helps for the same practical reason: it takes the job of tracking elapsed time out of your head.

Try 15/3 for a week. Lengthen it when the break starts feeling like an interruption rather than a relief, which is the signal that the interval is now shorter than your attention. This is a note about interval length, not medical advice.

How much focus time each schedule buys you

Longer blocks feel like more work and are not. This is arithmetic rather than a measurement: each row runs its own cycle through an eight hour day and stops when the next block would not fit whole.

The spread from best to worst is 63 minutes a day, which is less than an hour and far less than the difference between a good day and a bad one. Pick the schedule you will keep. The arithmetic is not going to decide it for you.

How to pick yours

1. Run 25/5 for a week without changing it

   You need a baseline, and one day is not one. Note which of the two complaints you have: the block ends too soon, or the block never starts.

2. Move in one direction only

   Too soon, go to 50/10. Never starts, go to 15/3. Changing the focus length and the break length and the long-break interval at once tells you nothing about which one helped.

3. Fix the break before you touch the work

   Most people who think 25 minutes is wrong are really finding that a 5 minute break is not enough to reset. Take the technique at its word and try 10 first, which is inside the range it actually specifies.

What the interval cannot fix

No length helps with a task you have not defined. The part of the technique that does the work is deciding what the next block is for before you start it, and every ratio in this article is downstream of that.

Sources checked on 30 August 2026: pomodoro.pomodorotechnique.com for the official timer's behaviour, the DeskTime blog post "Does the 52-17 rule really hold up?" of 3 October 2024 for the 2014, 2020 and 2021 ratios, the Wikipedia article on the Pomodoro Technique and the studies it cites for the research table, and each timer's own settings panel for the defaults. The two measured tables were run here, in Chrome 151 on macOS 26.6, on the same day.