The most famous number in time management started as a workaround. In the late 1980s Francesco Cirillo, a university student who could not make himself study, used a tomato-shaped kitchen timer to test a small question: could he work for two minutes without interruption? Twenty-five minutes came later, through trial and error. Cirillo arrived at it by experiment on one person facing one problem, not from attention research.

That does not make 25 wrong. It makes it a default, not a finding, and the controlled studies that exist on work-break schedules do not single it out. Below: what Cirillo actually specified, what researchers have measured since, and how to choose an interval on evidence instead of habit.

How Cirillo Arrived at 25

By his own account, the first experiment was two minutes. Secondary accounts describe him stretching the interval step by step, up to an hour, and finding that the longer blocks stopped holding his focus. He settled on 25. In the 1990s he was mentoring software teams, suggested the method to clients who laughed at it, and one team tried it anyway. That team’s results spread it through the software community, and it later reached a wider audience through his free guide and a 2018 book.

The rules are stricter than most timer apps suggest.

Read together, these rules describe a discipline against interruption and a method for calibrating estimates. The timer is a commitment device. The 25 is the smallest unit that made the discipline workable for the person who invented it.

25 minutes was never a finding. It was a calibration, and the only person it was calibrated for was Cirillo. worldtimedata

What Controlled Studies Found

Direct tests of Pomodoro-style schedules are recent, small, and mostly done on university students in a single study session. Three of them point in different directions.

Two details matter. The 2023 study used 24 and 6 minutes, not 25 and 5, and its authors said that, to their knowledge, the schedule had not been tested empirically before. In it, students who chose their own breaks studied in longer stretches and took longer breaks, which was associated with more fatigue and distraction, while the fixed schedules reached similar task completion in less time. The 2026 authors add that their performance measure may have been too insensitive to detect differences.

Across the three, none showed fixed schedules completing more work. The 2023 study found the same amount finished in less time. What the timer changes is mostly how people feel, and that effect flipped direction between studies.

The Break May Matter More Than the Block

The stronger evidence concerns breaks themselves. A 2022 meta-analysis in PLOS ONE pooled 22 samples (2,335 participants) on breaks of 10 minutes or less. Micro-breaks produced small but significant gains in vigor (d = 0.36) and reductions in fatigue (d = 0.35). The overall effect on performance was not significant (d = 0.16), and performance gains showed up only for tasks with low cognitive demand. Across studies, longer breaks were associated with larger performance benefits, and the authors suggest that recovering from highly depleting work may take more than 10 minutes.

Applied to Pomodoro, that points at the 5-minute break as the weaker element for demanding work: enough to feel fresher, not necessarily enough to restore performance. This is a comparison across different studies, not a randomized test of break lengths, so treat it as a lead.

The other study often cited in support of 25 minutes is Ariga and Lleras (2011, Cognition). In a monotonous 50-minute vigilance task, 84 participants’ performance declined, except in the group that twice, briefly, had to respond to digits they had memorized beforehand. Their performance did not decline. The authors read this as a momentary switch refreshing the task goal, so attention does not habituate to it. It says nothing about 25 minutes, and the diversion was a small task, not rest.

52/17 and the Problem With Magic Ratios

The best-known alternative is 52 minutes of work and 17 of rest. It comes from a 2014 analysis by the time-tracking app DeskTime, which looked at its top 10% of users by productivity rating, defined as the share of time spent in apps classified as productive. That is an observational snapshot. It shows what highly rated users did, not that a 17-minute break made them productive.

DeskTime’s own repeats of the study show the “ideal” moving: 80/17 by February 2020, then 112/26 by 2021, with most of the growth during the pandemic. The optimum more than doubled in work minutes, from 52 to 112. A rule that moves with working conditions is describing conditions, not biology.

What can be compared is the ratio and the number of restarts each scheme implies.

25/5 and 50/10 have the same work share. The difference is that the second interrupts a task half as often, which matters when a task has an expensive setup. The cycle length also has a practical side effect. A 25/5 cycle takes 30 minutes and stays aligned to the half hour. A 52/17 cycle takes 69 minutes, so starting at 9:00 puts the next block at 10:09, then 11:18, then 12:27. Any cycle that does not divide the 60-minute hour drifts against a day built from clock-aligned meetings.

How Many Blocks Fit a Workday

A full Pomodoro set of four blocks with its breaks takes 130 to 145 minutes and holds 100 minutes of timed work. Over a shift, that means 69% to 77% of the time can be timed focus, and only if nothing else interrupts the schedule. To estimate it, get the length of the shift from a work hours calculator, subtract meetings and lunch, and multiply the remainder by 0.69 to 0.77. An 8-hour shift with one hour of lunch leaves 420 minutes, which is about 290 to 320 minutes of timed focus. That is roughly 5 hours, not 8, before a single meeting.

Choosing an Interval by the Problem

The evidence supports matching the schedule to the kind of work more than adopting any single number. Each row below says how strong the support is.

Because the group-level studies disagree, the reliable test is your own. Cirillo’s method already contains the instrument: estimate blocks per task, record what you finish, review. Run the standard block for a week on a 25-minute timer, note how many estimated blocks each task actually took and how tired you are by late afternoon, then change one variable, the block or the break, using any timer set to the new length, and repeat. That kind of experiment, not a table of studies, is how 25 was chosen in the first place. Log each block as a recorded time interval from the timer instead of trusting how long it felt, since the two can differ.