Short answer: ultradian rhythms are real biological patterns. The stronger productivity claim, that everyone should work for 90 minutes and then rest for 20, is not established by the research reviewed here. Studies of waking performance have produced mixed findings, and finding a rhythm is a different question from testing a work schedule.
A 90-minute block can still be useful. Treat it as a way to organize a task, with permission to pause, rather than a deadline your biology requires you to reach.
The short version
- Ultradian means shorter than a day. It describes a category of biological rhythms, not one universal attention timer (Goh et al., 2019).
- Sleep cycles are not automatically work cycles. NIH guidance describes sleep cycles of roughly 80 to 100 minutes. That does not establish an equivalent limit for focused work (NHLBI, 2022).
- The waking-performance evidence is mixed. An early study reported task-performance oscillations, while a later study of 60 people found no significant 90-minute rhythm (Klein & Armitage, 1979; Neubauer & Freudenthaler, 1995).
- Break research supports a narrower claim. A meta-analysis found small improvements in energy and fatigue from short breaks, but no significant overall performance benefit. It did not establish a 90/20 schedule (Albulescu et al., 2022).
What is an ultradian rhythm?
An ultradian rhythm is a recurring biological pattern with a period shorter than 24 hours. Examples include sleep-stage cycles and patterns of hormone secretion. A circadian rhythm has a period of about 24 hours (Goh et al., 2019).
| Term | Approximate timescale | What it describes |
|---|---|---|
| Ultradian rhythm | Shorter than 24 hours | Repeated biological activity within a day; different processes can have different periods |
| Circadian rhythm | About 24 hours | The daily timing of biological functions |
| 90-minute work cycle | A chosen 90-minute block | A planning method; setting a timer does not measure a biological rhythm |
The singular phrase can be misleading. There is no reason to assume that every process described as ultradian shares the same period or reaches its peak at the same time. A review by Goh and colleagues emphasizes that many of these events have irregular timing; the authors prefer the term "episodic ultradian events" for the phenomena they discuss.
Where the 90-minute idea comes from
Sleep researcher Nathaniel Kleitman proposed a basic rest-activity cycle, often shortened to BRAC. His 1982 review argued that the pattern associated with alternating non-REM and REM sleep also had a counterpart during wakefulness (Kleitman, 1982).
That is the historical bridge between sleep-cycle research and advice about daytime work. It is a hypothesis about recurring activity, not an experiment comparing 90 minutes of writing plus 20 minutes of rest with other schedules.
The distinction matters because a measured biological period does not, by itself, tell you how long a work session should last. A scheduling recommendation also needs evidence about outcomes: what people finish, how accurate their work is, and how they feel afterward.
What the evidence supports
1. Sleep has repeating stages, but that does not set a work limit
The US National Heart, Lung, and Blood Institute describes a sleep cycle that restarts roughly every 80 to 100 minutes, usually occurring four to six times in a night (NHLBI, 2022).
What this does not show: this is information about sleep architecture. It does not test studying, coding, or office work, and it does not identify a mandatory rest point 90 minutes after you open your laptop.
2. Studies of waking performance have disagreed
In an early experiment, researchers measured verbal and spatial matching performance every 15 minutes for eight hours. They reported oscillations lasting 60 to 100 minutes, with the two task patterns out of phase (Klein & Armitage, 1979).
A later experiment tested 60 people every 10 minutes for nine hours. It measured sentence verification, ratings of alertness and mood, and heart period. The researchers found no significant 90-minute periodicity in those variables using their more conservative analysis (Neubauer & Freudenthaler, 1995).
| Study | What was measured | What it found | What it does not establish |
|---|---|---|---|
| Klein & Armitage (1979) | Verbal and spatial matching every 15 minutes for eight hours | 60- to 100-minute oscillations, out of phase between tasks | A shared 90-minute peak in all abilities or an optimal break schedule |
| Neubauer & Freudenthaler (1995) | Cognitive performance and other measures in 60 people over nine hours | No significant 90-minute rhythm under the reported conservative analysis | That all ultradian biology is absent |
| Búzás et al. (2024) | Recorded human physical activity and a mathematical model | Activity bursts organized at several timescales | That movement patterns identify the best duration for knowledge work |
The 1995 authors also reported that apparent rhythms depended on how the time series were processed and analyzed. That makes the choice of analysis part of the evidence, rather than a technical detail we can ignore.
What this does not show: neither cognitive experiment tested whether assigning people a 90/20 work-break schedule improved their output. The findings concern patterns during tasks, not the effectiveness of a productivity method.
3. Newer activity research does not settle the productivity question
Búzás and colleagues studied the organization of human physical activity. They described short activity bursts grouped into structures at longer timescales and developed a model reproducing important statistical features of the recordings (Búzás et al., 2024).
This is relevant to how activity unfolds during the day. But physical movement is not the same outcome as reading comprehension, useful code, or a finished report. The study should not be presented as a trial showing that everyone needs a break after 90 minutes.
4. Breaks can help without proving a 90-minute rule
A meta-analysis combined 22 independent study samples examining micro-breaks of 10 minutes or less. It found small improvements in vigor, or energy, and reductions in fatigue. The overall performance effect was not statistically significant (Albulescu et al., 2022).
| Outcome | Effect size | Studies / participants | Statistically significant? |
|---|---|---|---|
| Vigor (energy) | d = 0.36 | 9 studies, 913 people | Yes |
| Fatigue (lower) | d = 0.35 | 9 studies, 803 people | Yes |
| Overall performance | d = 0.16 | 15 studies, 1,132 people | No |
What this does not show: the review did not validate a 90-minute work interval or a 20-minute recovery period. Its break-length definition also means it cannot directly tell us how effective a 20-minute break would be.
What this research has not shown
The sources reviewed here do not establish:
- A universal 90-minute attention span for adults.
- That a 90/20 schedule beats 25/5, 50/10, or breaks chosen by the worker.
- That a slump at a particular time identifies the phase of an internal ultradian clock.
- That scheduling work around presumed ultradian cycles improves real-world output over weeks or months.
These are limits of this review, not a claim that every possible study has been ruled out. Biological observations, laboratory task results, and work-schedule trials answer different questions.
How to use this information
Use longer work blocks as an experiment in planning. You do not need to prove that you have synchronized with an internal clock to find a useful routine.
- Pick a task with a visible outcome. For example, outline one section of a report or investigate one reproducible software bug.
- Start with a manageable block. Use 25 or 50 minutes if that fits your current work. Try 90 when you have a task and a calendar window that suit it.
- Choose a break in advance. A 10- or 20-minute pause can be a practical scheduling choice. Neither number is presented here as a proven biological reset.
- Compare equal amounts of available time. Include breaks when comparing a longer block with shorter intervals. Record completed work, errors or rework, and fatigue from 1 to 5.
- Change one thing at a time. Try the alternative on several comparable days. Note interruptions and unusually poor sleep so they do not disappear from your interpretation.
- Pause when you need to. Reaching 90 minutes is not the objective. Finishing useful work at a sustainable pace is.
For example, reserve the same two-hour window on several days for similar writing tasks. Try your usual schedule on some days and a longer block with planned rest on others. Keep the task and time of day reasonably similar. This is an informal comparison, not a laboratory measurement of your rhythm.
You can try a longer window with the 90-minute timer, or use the 50/10 preset for shorter blocks with planned breaks. The countdown tracks time; it does not detect your biological state.
Ultradian rhythm vs Pomodoro: which should you use?
Pomodoro offers a defined routine with relatively frequent stopping points. A longer block leaves more time between planned interruptions. Choose according to the task, your obligations, and what you observe when you try each schedule.
If getting started is the main difficulty, a short commitment may feel more manageable. If a task needs a long setup and interruptions are costly, a longer window may be convenient. These are practical reasons to experiment, not findings that one interval matches everyone's biology.
For the evidence on shorter fixed breaks, see Does the Pomodoro Technique actually work?. You can also use the deep work timer and adjust the durations in Settings.
Frequently asked questions
Are ultradian rhythms scientifically proven? Biological processes with periods shorter than a day are well documented. A universal 90-minute cycle that dictates the best work schedule is a separate claim, with mixed evidence from waking-performance studies.
How long is an ultradian rhythm? The term covers periods shorter than 24 hours. It does not mean exactly 90 minutes. The duration depends on the biological process being discussed.
What is the difference between ultradian and circadian rhythms? Circadian rhythms operate on an approximately daily timescale. Ultradian rhythms recur more frequently within that day. Neither term alone specifies a work-break schedule.
Should I work for 90 minutes and rest for 20? You can try it as a schedule if it suits your task, but the sources reviewed here do not establish it as optimal. Include breaks in your comparison with other routines and pause earlier when needed.
How do I find my own ultradian rhythm? A log can help you notice when work feels easier or harder. It cannot establish an internal biological period on its own. Use it to adjust your routine without treating each dip as proof of a 90-minute cycle.
Is a 90-minute work cycle better than Pomodoro? This review does not establish a winner. Compare schedules using completed work, quality, and fatigue rather than how closely you followed a timer.
Sources
- Albulescu, P., Macsinga, I., Rusu, A., Sulea, C., Bodnaru, A., & Tulbure, B. T. (2022). "Give me a break!" A systematic review and meta-analysis on the efficacy of micro-breaks for increasing well-being and performance. PLOS ONE, 17(8), e0272460. Full text.
- Búzás, A., et al. (2024). Hierarchical organization of human physical activity. Scientific Reports, 14, 5981. Research repository record and published paper. DOI: 10.1038/s41598-024-56185-0.
- Goh, G. H., Maloney, S. K., Mark, P. J., & Blache, D. (2019). Episodic ultradian events—ultradian rhythms. Biology, 8(1), 15. Full text.
- Klein, R., & Armitage, R. (1979). Rhythms in human performance: 1 1/2-hour oscillations in cognitive style. Science, 204(4399), 1326–1328. PubMed record and abstract.
- Kleitman, N. (1982). Basic rest-activity cycle—22 years later. Sleep, 5(4), 311–317. Publisher summary.
- National Heart, Lung, and Blood Institute. (2022, March 24). How sleep works: Sleep phases and stages. NIH guidance.
- Neubauer, A. C., & Freudenthaler, H. H. (1995). Ultradian rhythms in cognitive performance: No evidence for a 1.5-h rhythm. Biological Psychology, 40(3), 281–298. PubMed record and abstract.
How we checked: consulted the open-access biology review and break meta-analysis, NIH guidance, and the available research abstracts and publisher summaries. This is a focused evidence review, not a systematic review; abstracts cannot support every detail of a study's methods.
Last reviewed: 17 October 2026. Found an error? Send a correction.