There is no science behind 25 minutes
Let us be straightforward about this. Twenty-five minutes is not an experimentally derived optimum. It is roughly what Cirillo's kitchen timer wound to and roughly what worked for him after some trial and error.
Research on sustained attention does not point to a single magic number. How long you can stay with a task depends on its difficulty, your interest in it, how you slept, the hour of the day, and substantial individual variation. Even the familiar claim that attention collapses after fifteen minutes of a lecture rests on much weaker evidence than its popularity suggests.
So why did 25 stick? Because it is practical rather than precise. It is short enough that starting requires no resolve, long enough to make visible progress, and it cuts a day into blocks that are easy to plan with. Four sessions plus breaks is roughly two hours, which is convenient arithmetic.
Set the length by warm-up cost
One question settles most of it: how long does it take you to get back into this task after stepping away?
Low warm-up cost - 20 to 25 minutes
Vocabulary drills, practice problems, filing, inbox triage. You can leave and return at almost no cost. Cutting these into short blocks is actively better: the spacing helps retention, and the break arrives before boredom accumulates.
High warm-up cost - 45 to 90 minutes
Writing code, working through a proof, reading a dense paper, drafting long prose. These require holding several pieces in working memory at once, and it usually takes ten to fifteen minutes before you are properly running. Stopping at 25 means stopping just as you arrive.
Fifty minutes on and ten off works well here, as does ninety on and twenty off. Long sessions are expensive, so two or three a day is a realistic ceiling.
Tasks you are avoiding - 10 to 15 minutes
When starting is the whole problem, make the offer small enough that refusing feels silly. Fifteen minutes is hard to argue with. You will usually continue past it, and if you do not, it still beats zero.
Why the break is only five minutes
Five minutes is not how long recovery takes. It is roughly the longest break most people reliably come back from. Decide to rest for fifteen minutes, start a video, and the afternoon is gone.
That makes the content of the break matter more than its length. Staying on a screen does not restore visual or attentional resources. Standing up, walking, looking out a window, drinking water, stretching - anything categorically different - does.
You do not need to be rigid about it. Ten minutes after a long session and twenty to thirty after every fourth, scaling with accumulated fatigue, is more natural than a fixed five.
Finding your own number
A week of observation beats any amount of theory. Run it like this.
- Spend three days on the default 25 and 5 to establish a baseline.
- At the end of each session write one line: did stopping feel like a loss, or were you already spent?
- If 'a loss' repeats, add five minutes. If 'already spent' repeats, subtract five.
- Expect different answers for different subjects or task types. Keep two or three presets rather than forcing one number on everything.