The Pomodoro Method: How Long to Work and How Long to Rest

What the planner returns, and the long break it leaves out

Five inputs, four outputs, and less arithmetic than you would expect. Total focus is the work interval times the number of sessions. The number of gaps is the number of sessions minus one, because there is no break after the last one. Long breaks go into as many of those gaps as there are completed cycles, minus one if the sessions divide evenly by the cycle length — which is the line that makes four sessions of four-per-cycle produce no long break at all. Every remaining gap gets a short break. Add it up and you have the total; the tool prints it as hours and minutes.

Run it across session counts and the pattern shows: one session, no breaks at all; four sessions, three short breaks and no long one; five sessions, one long break; eight sessions, still one; nine sessions, two. If you were expecting a long break at the end of every fourth session, that is not what you get, and the difference is not trivial — an afternoon planned as eight sessions gets a single fifteen-minute recovery in the middle and nothing else longer than five minutes.

Three smaller things the planner does without telling you. Total sessions is forced to at least one, so asking for zero returns a plan of one session, and 2.9 is floored to 2. Sessions per cycle is forced to at least one as well, so a zero there becomes a long break after every single session — 105 minutes of break on eight sessions instead of 45. And the total duration is assembled as whole hours plus rounded minutes, which produces "60 min" at 59.7 total minutes and "1 h 60 min" at 119.7. You will only see those if you type a fractional work interval, but if you do, that is why.

Where 25 and 5 come from

Francesco Cirillo devised the technique as a student in the late 1980s, timing himself with a tomato-shaped kitchen timer — pomodoro is Italian for tomato. His own presentation of it has always been broader than the interval: the full method is a planning system with daily planning, interruption handling and effort estimation, and the timer is the part that makes the rest enforceable. The 25 minutes was a length that worked for him and turned out to be teachable. It was never the winner of a comparison.

That matters because the internet has retrofitted a biology onto the number — ultradian rhythms, attention spans, a brain that empties every twenty-five minutes. None of that survives contact with the source. There is no published trial randomising people to 25/5 against, say, 40/10 or 50/10 on the same task and measuring output. The 52/17 figure that circulates alongside it is worse: it comes from a time-tracking company's analysis of its own users' logs, which is a description of what productive users happened to do, not a demonstration that doing it makes you productive.

What the research actually measured

The best-known laboratory study of interrupted work found something almost nobody quotes. Gloria Mark, Daniela Gudith and Ulrich Klocke had subjects answer email under three conditions and measured the time to finish. The uninterrupted group was the slowest: 22.77 minutes on average, against 20.31 for those interrupted by something related and 20.60 for something unrelated. There was no difference in the number of errors. People did not lose time to interruptions; they compensated by working faster. What they paid instead shows up on the workload scales: on a 1-to-20 measure, stress went from 6.92 uninterrupted to 9.46 interrupted, frustration from 4.73 to 6.63, effort from 9.50 to 11.04, and the uninterrupted group also wrote noticeably longer messages.

The other study everyone cites for breaks is Atsunori Ariga and Alejandro Lleras's 2011 paper in Cognition, and it is routinely misread. Their subjects did a visual vigilance task for about fifty minutes while holding digits in memory. The group that only recalled the digits at the very end showed the usual steep decline in performance; the group asked to recall them sporadically during the task did not decline. The authors' explanation is goal habituation: briefly deactivating the vigilance goal by switching tasks stops its activation from fading. Note what the "break" was — a short switch to a different mental task, not a rest, not a walk, not a coffee. The finding is real and it does not say that resting restores an attentional fuel tank; the paper explicitly argues against that view.

Put those two together and the honest summary is narrow. Interruptions are not primarily a time cost; they are a stress cost, and the compensating speed-up may itself cost you something the studies did not measure. Periodically letting go of the goal and coming back to it protects sustained attention on a repetitive task. Neither result licenses a specific number of minutes, and neither was run on anything resembling writing a report or debugging a program.

Why a fixed interval works anyway

Because it makes starting cheap. The hard part of most work is not the twenty-third minute; it is the first, and the decision the interval removes is the one you were going to lose the afternoon to. Twenty-five minutes is small enough to be agreed to by someone who does not want to begin, and long enough that beginning is not pointless. The same is true of thirty, or of twenty. The mechanism is commitment, not dosage — which is exactly why the exact figure has never needed to be defended.

There is a second mechanism, and it is the one the planner serves. A fixed interval makes work countable. Once a session is a unit, an afternoon has a size, and "finish the chapter" becomes "six sessions", which is a claim you can be wrong about and then correct. That is Cirillo's estimation step, and it is the part of the method that survives longest in practice — people stop watching the clock within weeks and keep counting the blocks for years.

Where it fails: work with a long ramp-up

Some work has a cost of entry that a twenty-five minute block cannot amortise. Reading yourself back into a proof, reconstructing the state of a system you were debugging yesterday, holding six threads of an argument at once — for these, the first ten or fifteen minutes buy nothing but the ability to start being useful, and a timer that fires at twenty-five ends the session shortly after it began. Worse, the interruption lands exactly where it hurts: at the point where the loaded state is most expensive to rebuild.

The fix is to change the numbers, and the planner is happy to take them. Ninety minutes of work, a twenty-minute short break, a thirty-minute long break, two sessions per cycle and four sessions gives 360 minutes of focus, 70 of break, one long break and 7 h 10 min in total — 83.7 % focus, a slightly denser day than the default despite the longer breaks, because there are fewer gaps to fill. Fifty minutes with a ten-minute break, two per cycle over six sessions gives 300 focus, 90 break, two long breaks, 6 h 30 min. Pick the interval that fits the ramp-up of the work, not the interval a kitchen timer happened to have.

One rule survives every variation, and it is the one people break: a break is only a break if it is not another screen. Cirillo's own instruction is to step away, and the mechanism the vigilance research points to is releasing the goal — neither of which happens when the timer rings and you open a feed. That is the method's rule rather than a measured finding, and it is stated here as such, because no study was found that compares five minutes of scrolling with five minutes of standing at a window. But if you keep the interval and lose the break, you have kept the part that was never doing the work.

Frequently asked questions

Is 25 minutes the right length?

There is no published trial that establishes it, and none that establishes any competing figure either. Twenty-five is Cirillo's own working length from the late 1980s, and it spread because it is teachable, not because it beat a control condition. What is defensible is the shape of the choice: pick a block short enough that you will actually start it and long enough to clear the cost of getting into the work. For email and admin that can be fifteen; for anything with a long ramp-up, ninety is more honest. The planner will cost out any of them.

Why does eight sessions give only one long break?

Because the tool places a long break after each completed cycle except the final one, on the reasoning that a break at the very end of the plan is not a break — you have finished. Eight sessions at four per cycle are two completed cycles, so two long breaks minus the trailing one leaves one. Four sessions are one completed cycle and leave none at all: three short breaks and nothing longer. If you want a long break at the end of the day, add it to your own schedule; the planner will not count it.

Does this tool actually run a timer?

No. Despite the name, there is no clock, no countdown, no alarm and no notification — it is a planner, and everything it returns is known before you start. That is not a criticism so much as a scope: the question it answers is "how long will this take me end to end, and how much of that is actually work", which is the question you have before an afternoon rather than during one. Use a phone timer, an egg timer or the mechanical tomato for the running part.

Why did it show me "1 h 60 min"?

Because the total duration is assembled as whole hours plus separately rounded minutes, and when the remainder is 59.5 minutes or more, rounding it produces 60 rather than carrying into the hour. You will only trigger it with a fractional work interval — 119.7 total minutes prints as 1 h 60 min, and 59.7 prints as 60 min. The number is right; only the writing is wrong. Read the focus and break figures instead, or type whole minutes.

What should I actually do in a five-minute break?

Anything that is not the thing you were doing and not another feed. Cirillo's instruction is to step away; the vigilance research points to briefly letting go of the goal rather than to rest as such. Standing up, water, a window, a short walk to another room all qualify; opening a messaging app does not, because it substitutes one demanding task for another and adds the resumption cost of a second thread. This is the method's rule, not a measured finding — no study was found comparing five minutes of scrolling with five minutes away from the desk — but it is the rule that the people who abandon the method break first.