Kleitman's Ultradian Rhythm: The 90-Minute Work Cycle

Nathaniel Kleitman was the University of Chicago physiologist who is widely considered the founder of modern sleep research. In the 1950s and 1960s, his lab discovered REM sleep and mapped the structure of the sleep cycle. His most consequential extension to waking life was the Basic Rest-Activity Cycle, or BRAC, which he proposed in 1969. The BRAC is the observation that the same approximately 90-minute rhythm that drives sleep stages at night continues during waking hours, expressing as alternating periods of high and low alertness across the day.

The 90-minute interval is now widely cited in deep-work research and elite-performance research. K. Anders Ericsson's work on the practice habits of world-class performers (musicians, athletes, chess players) found that top practitioners consistently cap focused practice sessions at around 90 minutes per session, with three to five sessions per day separated by long breaks. The constraint is not motivation; it is the cognitive recovery cycle Kleitman first described.

This post covers what the science actually says, why 90 minutes works for deep cognitive work, and how to apply the rhythm without overinterpreting it. For shorter-interval alternatives, see our deep-dives on the 25-minute Pomodoro and the DeskTime 52/17 study.

Who Kleitman was and what he discovered

Kleitman ran the world's first sleep research laboratory at the University of Chicago. His 1939 book Sleep and Wakefulness was the foundational text in the field for decades. In 1953, he and his graduate student Eugene Aserinsky discovered REM (rapid eye movement) sleep, which mapped the sleep cycle's distinct stages for the first time.

The sleep stages cycle through approximately every 90 minutes during the night. A typical 8-hour sleep produces 4 to 5 full cycles, each containing light sleep, deep sleep, and REM. The cycle length is not exactly 90 minutes; individual variation runs from 80 to 120 minutes. The mean is approximately 90.

Kleitman's BRAC hypothesis (1969) was that the same 90-minute pattern continues during waking hours, expressing as alternating peaks and troughs of alertness, cognitive capacity, and physiological arousal. The hypothesis was based on observations of sleep patterns plus monitored alertness data during the day. It was not perfectly proven at the time and remains an active research area, but the directional finding (a roughly 90-minute waking cycle) is well-supported.

What the BRAC actually is

The waking BRAC has three observable features.

• Roughly 90-minute period. The cycle from one alertness peak to the next is approximately 90 minutes, with individual variation. Some people run 80-minute cycles, others 110.
• High-alertness phase (roughly 60 to 90 minutes). The portion of the cycle where focus, attention, and cognitive capacity are above average. This is when deep work is easiest.
• Low-alertness phase (roughly 20 to 30 minutes). The portion where alertness drops, fatigue surfaces, and continued focused work produces lower-quality output. This is when the brain wants a break, whether you take one or not.

The cycle is roughly synchronized to circadian rhythm but not identical. Most people experience their strongest focus peaks in the morning (around 9 to 11 AM) and another peak in the afternoon (around 2 to 4 PM, after the post-lunch dip). The exact timing varies by chronotype, with morning people peaking earlier and night owls later.

How K. Anders Ericsson's research extended this to expert performance

Ericsson was the cognitive psychologist who developed the deliberate practice framework. His research compared the practice habits of top-tier performers (concert violinists, chess grandmasters, elite athletes) against good but non-elite performers in the same fields.

One of his consistent findings was that top performers cap focused practice sessions at around 90 minutes per session. They do not practice for 4 hours straight. They do three to five separate 90-minute sessions per day with significant breaks between them. The pattern held across domains, even though the specific practice content differed dramatically (a violinist practicing scales versus a chess player analyzing positions).

Ericsson's interpretation: 90 minutes is the upper limit on the kind of deeply focused, attention-demanding practice that produces expert-level skill development. Beyond 90 minutes, the quality of attention drops to the point where additional time produces diminishing or even negative returns. This is consistent with Kleitman's BRAC: the high-alertness phase ends, the low-alertness phase begins, and pushing through the trough produces fatigued practice that does not move the needle.

The implication for knowledge work is direct. If 90 minutes is the cap for elite practitioners' focused work, it is at minimum a useful upper bound for sustained deep cognitive work in any field.

Why 90 minutes works for deep work specifically

Three mechanisms make the 90-minute cycle particularly well-suited for deep cognitive work.

1. Long enough to ride the alertness peak

The 60 to 90-minute high-alertness phase is exactly long enough to use most or all of it for productive work. Shorter intervals (25-minute Pomodoros) end before the peak does, leaving capacity unused. Longer intervals (2-hour blocks) extend past the peak into the low-alertness trough, where additional time produces fatigued output.

2. Aligned with deep-work warmup costs

Deep work requires 5 to 15 minutes of mental warmup before useful output begins. A 90-minute block has 75 to 85 minutes of post-warmup productive time. The warmup-to-output ratio is favorable; you spend less time per block re-loading context than you would in shorter cycles.

3. Matches the body's natural recovery rhythm

Taking the break at the natural alertness trough means the rest is genuinely restorative. The body and brain are already cycling toward recovery; you are working with the rhythm rather than against it. Breaks taken during high-alertness phases (the failure mode of strict 25-minute Pomodoros) interrupt productive output to enforce a break the body did not need yet.

How to apply the 90-minute cycle to your work

For deep work in single sessions

1. Set a 90-minute work timer at the start of a focus block.
2. Use a forced-break tool that fires automatically. Doggy Break has a 90-minute preset specifically for this rhythm and labels it with the BRAC research in the settings panel.
3. Take a 15 to 20-minute break at the end of the cycle. Get up, move, look at something far away, drink water.
4. Run another cycle, or stop for the day if you have completed three to five cycles already.

For elite-performance practice

Ericsson's research suggests three to five 90-minute sessions per day with at least 30 to 60 minutes between sessions. Not all sessions need to be the same task; you can do morning research, afternoon writing, evening admin, with the cycles separating by content.

For mixed work days

Most knowledge workers do not have the autonomy to structure days as three to five pure 90-minute sessions. The realistic adaptation is to identify two protected 90-minute deep-work blocks per day (morning and afternoon, ideally aligned with your personal alertness peaks) and accept that the rest of the day will be shorter cycles or meetings. The two protected blocks are the ones that should follow the ultradian rhythm.

Benefits of working with the ultradian rhythm

1. More work done in fewer hours

Riding the alertness peak produces more output per minute than working through the trough. Two 90-minute peak-aligned blocks often produce more useful work than three 60-minute blocks that include trough periods.

2. Genuine recovery between blocks

Taking the break at the alertness trough means the rest is restorative rather than perfunctory. The 15 to 20-minute break is long enough to fully reset; the 5-minute Pomodoro break often is not.

3. Better fit for elite-performance practice

If you are doing skill-development practice (learning a language, music, sport, technical skill), the Ericsson 90-minute cap maximizes the deliberate-practice quality of each session. Longer sessions reduce per-minute skill acquisition.

4. Aligned with circadian preferences

The ultradian rhythm is layered on top of the circadian rhythm. Working with both (90-minute blocks during your circadian alertness peaks) produces compounding benefits over working with only one.

5. Less mental fatigue at end of day

Three 90-minute blocks plus breaks add up to roughly 5 hours of work. The remaining work day handles meetings, admin, email. The peak-time deep work is done by 1 PM, leaving the afternoon for lower-cognitive tasks. This pattern produces less overall fatigue than spreading deep work evenly across the full day.

Frequently Asked Questions

Is the BRAC the same for everyone?

No. Individual cycle lengths vary from approximately 80 to 120 minutes. The 90-minute number is the population mean. To find your personal rhythm, work in 90-minute blocks for two weeks and notice when your alertness consistently drops. If it drops at 75 minutes, your cycle is shorter; if you can sustain past 90, it is longer. Adjust accordingly.

Does the rhythm change with age?

Slightly. Older adults tend to have shorter cycles and earlier alertness peaks (which is part of why elderly people often wake earlier). The variation is real but modest; 90 minutes is still a reasonable starting point at any age.

Can I train myself to have longer focus cycles?

The BRAC is biological and fairly stable. You can train yourself to push through the alertness trough (this is what cramming feels like), but you cannot train yourself to have a different cycle length. The smarter strategy is to align work with the rhythm rather than fight it.

How does sleep quality affect the daytime BRAC?

Significantly. Poor sleep produces shallower alertness peaks and deeper troughs, making 90-minute cycles harder to sustain. The sleep rhythm and waking rhythm are aspects of the same biology. If 90-minute blocks consistently fail for you, check sleep quality before changing the work cycle.

Is caffeine compatible with the BRAC?

Yes, with timing matters. Caffeine peaks in blood plasma 30 to 60 minutes after consumption. A coffee at the start of a 90-minute block reaches peak around the middle of the block, supporting the alertness peak. Coffee taken too late (right before the cycle ends) extends alertness into the natural trough and disrupts the recovery break. The rhythm works best when caffeine is timed to the block start.

Can I do back-to-back 90-minute blocks without a long break?

Generally no. The 15 to 20-minute break between blocks is what allows the alertness trough to genuinely recover. Skipping it pushes you into the next cycle's high-alertness phase already partially depleted, which produces lower-quality output throughout the second block. Ericsson's data on elite practitioners specifically shows real recovery between sessions, not just brief pauses.

How does the 90-minute cycle interact with the NIOSH hourly break recommendation?

They are not contradictory. The NIOSH hourly break is an ergonomic safety floor for screen-related strain (eye, posture, circulatory). The 90-minute cycle is a cognitive optimum for deep focus. The reconciliation: take a 30-second postural micro-break at the 60-minute mark inside a 90-minute block, then take the full 15 to 20-minute break at cycle end. Both recommendations satisfied. We covered this in detail in the NIOSH research deep-dive.