ManageTime

Article

How long should a focus block be, and when should you break it

No single block length is correct, and the strongest evidence points at a decision rather than a number. A 2025 meta-analysis of 68 studies found that mind-wandering rises steadily the longer you stay on one task, so the length that works is the one ending before your attention already has. Match it to the work. This article covers what that research actually measured, why the case for breaking when you feel like it is weaker than it sounds, and how to pick a length for the task in front of you.

How long can you actually hold attention on one task?

Less time than the block on your calendar assumes. In 2025 Zanesco and colleagues published an individual-participant meta-analysis in Psychological Bulletin, pooling 68 research reports, more than 10,000 people and close to half a million answers to probes that interrupted a task to ask what the person was thinking about. Mind-wandering rose steadily across those probes. The longer a task ran, the more of it was spent somewhere else.

That is a slope, not a cliff. Attention does not hold flat for 25 minutes and then collapse; it drifts from the first minute and keeps drifting. The paper cites earlier estimates putting mind-wandering at 30 to 50 percent of waking moments, so a block in which your mind never left was never on offer.

A block contains that decline. It does not cure it.

What the meta-analysis cannot hand you is a number for your own desk. It mapped the shape of the decline, not the minute it turns expensive, and it mapped it in laboratory tasks with probes firing at set intervals rather than across an afternoon of real work. What transfers is the direction. Whatever length you pick, the last quarter of it is worth less to you than the first.

That has one consequence worth acting on this week. If the tail of a block is its weakest part, a block ending while you are still working beats one that runs until you stop. Finishing early costs you a few minutes. Running long costs you those minutes plus the time it takes to notice they went nowhere.

Is 25 minutes or 90 minutes better for hard work?

Neither number came out of a study, and none of the research behind this article compared them. The 25 minutes belongs to Francesco Cirillo, whose Pomodoro Technique pairs the sprint with a 3 to 5 minute break and a written record of every finished round. The 90 to 180 minutes belongs to Cal Newport's Deep Work, where the length is set by how long a hard problem takes to get properly into your head. Both are arguments their authors make, not measurements anyone reported.

What separates them at your desk is restart cost. Some work has to be loaded before it can be done: a contract you are halfway through, a bug you were three files deep into, an argument you had almost finished making. Put that down at 25 minutes and you pay the loading again at minute 26. Other work picks up cold. Clearing a queue or filling in a form charges you nothing for coming back.

So the useful question is not which number is right. Ask what your task costs to restart.

The second input is the week you have rather than the one you would design. A 90-minute block does not exist on a day holding a stand-up at ten and a call at eleven, and the focus pillar treats that as a scheduling problem rather than an effort problem. Waiting for a cleaner day is how the hard task keeps sliding into the next one.

The right length is the one your week can supply.

A workable middle is to start from the longest stretch your ordinary week reliably delivers, then set the block just under it. Timeboxing is the method built around fixing that length before you begin, and its own first instruction is to define what done means for the box. This does not fit everyone. If your calendar is written mostly by other people, no block length will hold until the calendar changes, and that is a conversation rather than a technique.

Should you break when the timer says so or when you lose focus?

Break when you feel your attention go is the usual advice, and the study that tested it most directly did not find that it mattered. Patel and colleagues ran five experiments on sustained attention, published in the Journal of Experimental Psychology: Applied in 2024. In the first two, breaks gave a small local benefit, and breaks the participants chose for themselves were no better than breaks the experimenter imposed on them.

The break research is also thinner than its reputation. Albulescu and colleagues pooled 19 studies covering 22 samples and 2,335 people in PLOS ONE in 2022. Micro-breaks lifted vigor and reduced fatigue, both small but statistically reliable effects. On overall performance the effect was smaller and did not reach statistical significance, and the performance benefit that did appear showed up only on the less cognitively demanding tasks, which is not where your hard work lives.

A short break makes you feel better. Whether it makes you work better is unsettled.

Two findings there do change what you do. Longer breaks predicted a bigger performance boost than shorter ones, and the authors suggest that recovering from heavily depleting work may need more than ten minutes. That puts the Pomodoro's 3 to 5 minutes at the short end of what has been tested. It is not a fault of the method; Cirillo's break exists to keep a sprint repeatable, and his technique is a tracking system before it is a rest schedule. After a long block on something hard, give yourself more room than a sprint gets.

The finding worth borrowing sits in the second half of the Patel experiments. When participants controlled the onset of each item rather than the timing of their breaks, they outperformed people working at a fixed rate, and they outperformed people handed the exact schedule a self-pacing participant had chosen.

Pace beat schedule. Deciding as you went was the part that worked, not the timing itself.

Inside a block that means holding the order loosely: take on the next demanding piece when you have the attention for it, and leave the mechanical work for when you do not.

Two limits come attached. Self-pacing did not remove the drop in vigilance that comes with fatigue, so it slows nothing down on its own. Its benefit also shrank when attention was split across two tasks, which is the condition a block with your chat window open is already in. These were laboratory attention tasks measured in minutes, not an afternoon at your desk.

How does your working style change the answer?

None of these studies asked how people organise a week, so what follows is this site's reading rather than a research finding. The four working styles describe how you already operate when nobody is supervising, and block length is one of the places that shows up fastest. Take the profile that sounds like your ordinary week.

If you build the week in advance, the Architect's pattern, your risk is not the length. It is treating a longer block as automatically better. Set the end before you start, put the work you cannot re-enter cheaply inside it, and let the block finish on time even on the days it was going well. Deep Work is the method that fits. Stopping is the part people drop.

Burst workers, closest to the Sprinter here, already have their unit at 25 minutes. The trap is stacking. Four sprints run back to back with the breaks skipped is a 100-minute block with a timer on it, and the Pomodoro Technique counts the break as part of the method rather than a reward for finishing.

A week that rarely repeats, which is the Improviser's situation, gives you the length rather than letting you choose it. Take what your calendar actually delivers. A 40-minute block that happens on Tuesday beats a 90-minute one waiting for a quiet week that keeps not arriving.

The Visionary's risk is different again: a block with no finish line inside it. Work described as make progress on the strategy can absorb 180 minutes and still be open at the end, so borrow Timeboxing's first instruction and write down what done means before the clock starts.

None of that is a length. It is a way of choosing one.

One test, then, rather than a new system. Pick your next hard task, set a length you are confident you can finish, and write one line afterwards: did the last quarter of that block still produce anything. Two weeks of those lines will tell you more about your own length than any number here. If you are not sure which of these is yours, the working style assessment takes a few minutes.

Frequently asked questions

How long should a focus block be?

Long enough to cover what your task costs to restart, and short enough that you end it while you are still working. No study names a correct length. The evidence gives a direction instead: mind-wandering rises the longer a task runs, so the end of any block is its weakest part. Set the length from the longest stretch your week reliably delivers, then stop on time.

Is 25 minutes or 90 minutes better for deep work?

Neither figure comes from research: 25 minutes is Cirillo's design, and 90 to 180 minutes is Newport's range for deep work. Choose by what your task costs to reload. A contract you are halfway through, or a bug three files deep, charges that cost every time you come back, and a long block pays it once. Work you can pick up cold does not need one.

Should you break when the timer says so or when you lose focus?

It makes little difference: in a 2024 set of experiments, self-chosen breaks were no better than ones the experimenter imposed. What helped in those same experiments was controlling the pace of the work itself, moment to moment, rather than the timing of breaks. So keep a scheduled break if you have one. Spend the control you do have on deciding when to take the next hard piece.

Why does my focus block stop working after the first hour?

Nothing breaks at the hour mark: the decline runs from the first minutes, and it only grows large enough to feel once the block has been going a while. A 2025 meta-analysis of 68 research reports and close to half a million probes found mind-wandering rising steadily with time on task. The curve has no step in it anywhere. Treat the fade as what a long block does, and end the block before it rather than after.

Sources