90-Minute Focus Blocks: Structure That Holds

A 90-minute focus block is one uninterrupted stretch of work on a single outcome, followed by a real break of roughly 15 to 30 minutes. The length is worth using, but not for the reason it is usually sold. There is no established 90-minute clock ticking in your waking brain. The figure was borrowed from sleep research, where the average cycle genuinely does run near 90 minutes, and carried across to daytime attention where the evidence is much thinner. What makes the block work is its internal shape: the first fifteen minutes are warm-up, the middle fifty-five are the actual work, and the last twenty are a fade you should plan for rather than fight.

That distinction matters more than it sounds. If you believe 90 minutes is a biological unit, you will treat a block that ends at 62 minutes as a failure. If you understand it as a container with a known shape, you will start structuring the inside of it, which is where all the available gains actually are.

What is a 90-minute focus block?

The rules are short. One outcome, not a list. No inbox, no chat, no phone in the visual field. A timer running so that you are not the thing keeping time. And a hard stop at the end followed by a break you actually take.

The reason it is not simply "a pomodoro but longer" is the warm-up. A 25-minute block spends a meaningful fraction of itself getting started, so the technique compensates by making starting easy and repeating often. A 90-minute block pays the startup cost once and then has close to an hour of usable depth on the other side of it. That is the whole trade: fewer, more expensive starts in exchange for longer runs of continuous thought.

Which also tells you what the block is for. Writing, coding, design, analysis, studying something genuinely difficult: work where you have to hold a lot of structure in your head at once and where being interrupted costs you the structure. It is a poor fit for email, admin, and anything you can do in five-minute pieces without losing your place.

Where does the 90-minute figure actually come from?

Two sources, and only one of them is solid.

The solid one is sleep. Human sleep moves through repeating stage cycles across the night, and the average cycle length is commonly quoted at about 90 minutes. The Sleep Foundation's breakdown of sleep stages puts the first cycle in the range of 70 to 100 minutes with later cycles typically running 90 to 120, and notes that cycles vary from person to person and night to night. The NCBI Bookshelf's StatPearls chapter on sleep physiology gives a similar picture, describing a complete cycle as roughly 90 to 110 minutes with four to six of them a night. So "about 90 minutes" is a real average in sleep, with real spread around it.

The shakier source is the leap from there to waking attention. The sleep researcher Nathaniel Kleitman, working outward from the discovery of REM sleep, proposed that the same rhythm continues through the day. The idea is usually called the basic rest-activity cycle, or BRAC. It is a genuine hypothesis with genuine research behind it, but it is looser than the productivity literature implies. A 2009 Neuroscience paper on the physiology of the basic rest-activity cycle describes it as a 1 to 2 hour ultradian periodicity in alertness and behavioural activity, with episodes occurring at variable intervals and behaving more like discrete events than a clean oscillation. A 1 to 2 hour band with variable intervals is a very different object from "your brain resets every 90 minutes."

And when researchers have gone looking specifically for a 90-minute rhythm in cognitive performance, it has not reliably shown up. One careful attempt, Neubauer and Freudenthaler's 1995 study in Biological Psychology, tested sixty participants every ten minutes across nine hours on a sentence-verification task while tracking heart rate, alertness and mood, and reported no significant 90-minute periodicity in any of those variables. The authors suggested that some earlier positive findings may have come from statistical methods that were not conservative enough.

So the honest summary is this. Attention is not flat across the day, energy does move in waves, and those waves are real enough to plan around. The specific number 90 is an average imported from sleep, and your own period, if you have a stable one at all, may sit anywhere in a wide band. Treat 90 as a sensible default container, not as a measurement of you. We reached a similar conclusion about a very differently sourced number in the piece on the 52/17 work-break ratio, and about the 25-minute block in what research actually says about the pomodoro technique. The pattern repeats because the underlying situation repeats: the block is useful, the specific figure is a convention.

Why do the first fifteen minutes feel wasted?

Because they are not the work. They are the cost of arriving at the work, and every long block pays it.

Three things are happening in that window. You are reloading context: where the file was, what the argument was, what you decided last time and why. You are shedding residue from whatever you were doing five minutes ago, which is why a block that starts straight after a meeting is slower to bite than one that starts after a walk. And you are lowering your own tolerance for interruption, which is a gradual process rather than a switch.

The practical consequence is that a 25-minute block spends a large share of itself on warm-up, and a 90-minute block spends a small one. That is the actual argument for the longer block, and it is a much better argument than ultradian rhythms.

It also means you should stop treating the warm-up as a personal failing. Fifteen slow minutes at the top of a block is normal. What you can control is making them shorter and cheaper: leave a note at the end of the previous block saying exactly where to resume, start with a small mechanical task inside the real work rather than a decision, and never open the block with your inbox.

What does the inside of a 90-minute block look like?

Here is the shape in detail. The minute marks are not sacred, but the phases are, and knowing which phase you are in changes what you should be doing.

Minutes	Phase	What is happening	What to do
0 to 3	Setup	Nothing productive yet	Close everything not needed. Phone out of sight. Start the timer before you start the work, not after.
3 to 8	Reload	Rebuilding context	Read your handoff note from last time. Do not re-plan. Do not reopen the decision you already made.
8 to 15	Traction	Warm-up ending	Do a small mechanical piece of the real task: fix the paragraph you left broken, write the failing test, redraw the sketch. Momentum, not quality.
15 to 40	Work, first half	The block is now paying	Head down on the one outcome. Ignore every impulse to look something up unless the work stops without it. Keep a scratch list of tangents.
40 to 55	Peak	Usually the best stretch	Do the hardest thinking here. If there is one judgement call in this block, make it now.
55 to 70	Work, second half	Still productive, quality drifting	Keep going, but move toward execution over invention. Good time for the parts you already know how to do.
70 to 80	Fade	Attention thinning noticeably	Stop starting new threads. Finish the piece in front of you rather than opening the next one.
80 to 88	Close out	Winding down deliberately	Tidy the workspace of the work: save, commit, name the file properly, clear the scratch list.
88 to 90	Handoff note	The most valuable two minutes	Write two or three lines to your next self: where you stopped, the next concrete action, and the one thing you were unsure about.

The handoff note is the part people skip and the part that pays the most. It is what turns tomorrow's fifteen-minute warm-up into a five-minute one. Write it as an instruction, not a summary. "Next: rewrite the second section to lead with the objection" is useful. "Worked on section two" is not.

And if you hit real flow at minute 85, do not stop. A timer is a tool for starting and for preventing drift, not a supervisor. The rule that keeps this from becoming an excuse is a simple one: you may run over, but you may not skip the handoff note and you may not skip the break. Run to a natural seam, whatever the clock says, then write the two lines and take the full break anyway. What you must not do is let an extra ten minutes of flow quietly become three hours without a break, which is the failure mode that makes people abandon the whole method a week later.

The inverse case matters too. If you have been staring at the same sentence since minute 55 and nothing is coming, ending early is a legitimate outcome. Write the handoff note, take the break, and start a fresh block. A block that ends at 62 minutes with a clear note beats one that limps to 90 and leaves you nothing to resume from.

How many blocks fit in a real working day?

Fewer than the arithmetic suggests. An eight-hour day looks like it holds four 90-minute blocks plus breaks and change, and almost nobody gets four.

Two blocks a day is a good, sustainable, genuinely productive target for most people with a normal calendar. Three is a strong day. Four happens on days with nothing scheduled, and it is worth noticing that the fourth is usually the weakest one. There is no research establishing a hard ceiling on daily deep work, and anyone quoting you a precise number is guessing, but the practical constraint is not mysterious: meetings fragment the day, and a 90-minute block needs a 90-minute hole to sit in. A calendar with a 10am and a 2pm meeting has exactly one clean slot in it, whatever the total free hours add up to.

Which leads to the only scheduling rule that reliably helps. Place the first block before the day can get hold of you, ideally as the first thing you do, and defend it harder than anything else on the calendar. First blocks succeed at a much higher rate than afternoon ones, mostly because nothing has had a chance to interrupt them yet.

What should the break after a block look like?

Longer than a pomodoro break, and a genuine change of state.

Fifteen to thirty minutes is the usual range, and the reason it is longer than a five-minute break is proportionality: you spent ninety minutes in one posture, one visual distance and one goal, and undoing all three takes more than a stretch at the desk. Get up, get your eyes off a screen and onto something far away, move, and get outside if you can.

What does not count is a break spent on the same screen. Checking chat is a task switch rather than recovery, and it reliably imports the next problem into your head before the current one has been put down. The other reliable failure is the untimed break, which is how fifteen minutes becomes fifty. Time the break as strictly as you time the work. The detail on what makes a longer break restore anything is in the 52/17 post, which goes through the break research properly.

When is 90 minutes the wrong length?

Often, and there is no shame in it.

It is wrong when your calendar is chopped into 30 and 60 minute pieces, because a block you cannot fit is a block you will not run. It is wrong for interrupt-driven work, on-call rotations, support queues and anything where being unreachable for ninety minutes is a real cost to someone else. It is wrong when starting is your actual problem, because the hardest thing about a 90-minute block is agreeing to begin it, and a 25-minute commitment gets you working while a 90-minute one gets you cleaning the kitchen. And it is wrong for the sort of work that does not have a warm-up to amortise, which is most administrative tasks.

In all of those cases the shorter block is not a compromise, it is the correct tool. A 45-minute block is a good middle setting: long enough to clear the warm-up and get real depth, short enough to fit between meetings and to commit to on a low-energy morning.

One last thing worth saying plainly. If you cannot concentrate for any useful length of time, and that is new for you, or you are exhausted every afternoon no matter how the day is arranged, block length is not the variable you are looking for. Persistent fatigue and a change in your ability to concentrate are worth raising with a doctor rather than solving with a timer. This is scheduling craft, not medical advice, and interval tuning is for a system that is basically working.

Run one block today rather than reading a second article about them. Open the countdown timer, set it to 90 minutes, write down the single outcome for the block, and start it before you start the work. It counts from the clock rather than from ticks, so it stays accurate if the tab goes quiet in the background while you are in the document. If 90 feels like too big a commitment for a first attempt, the one-click 45-minute timer is an honest place to begin and you can lengthen it once the shape is familiar. Every timer here is free, with no account and nothing to install.

FAQ

Are 90-minute work blocks better than pomodoros?

Neither is better in general, and no study establishes one length as optimal. They solve different problems. A 25-minute pomodoro minimises the cost of starting, which makes it the right tool for procrastination, fragmented calendars and interrupt-heavy jobs. A 90-minute block minimises the number of warm-ups you pay for, which makes it the right tool for writing, coding, analysis and other work with a long ramp. If your problem is beginning, go shorter. If your problem is being pulled out of depth, go longer.

How many 90-minute blocks can you do in a day?

Two is a realistic and sustainable target for most people with meetings in their calendar, three is a good day, and four generally requires a day with nothing else in it. The binding constraint is usually the calendar rather than your attention, because a 90-minute block needs a 90-minute clear hole, and a day with two scattered meetings may only contain one. Judge the day on completed blocks rather than hours logged, and protect the first block of the morning above the others.

What break do you need after 90 minutes?

Roughly 15 to 30 minutes, and it should change your state rather than just your task. Stand up, leave the desk, look at something at a distance, move, and get daylight if you can. A break spent on the same screen in a different tab is a task switch, not recovery. Time it, because the most common way a good break fails is by quietly running to fifty minutes.

Should you stop mid-sentence when the timer ends?

No. Run to a natural seam, then stop. The value of stopping at a known point is that you leave yourself something to resume from, which is why the last two minutes of the block should go on a short handoff note: where you stopped, the next concrete action, and the thing you were unsure about. Stopping mid-sentence saves nothing and costs you the note. Running an extra ten minutes to finish a thought is fine. Running an extra ninety and skipping the break is how people abandon the method.