Ultradian Rhythm: What the 90-Minute Cycle Research Actually Shows, and How to Block a Day Around It Honestly

The 90-minute productivity cycle is a sleep finding wearing a daytime costume.

Operators like the idea. Work in 90-minute blocks, rest, repeat, because the body runs on a 90-minute rhythm and fighting biology is a losing trade. It is tidy, it is quotable, and it has a number in it.

The number is real. It belongs to the night. Whether it governs your afternoon is a separate question, and the research answers it with far less confidence than the productivity books do.

This article separates what is established about ultradian rhythms from what is extrapolated, states the honest conclusion about work-block length, and gives the Apex way of blocking a day, along with a one-week experiment to find the block length that fits you rather than the one that fits the slogan.

What ultradian means

Circadian rhythms run on roughly a day. Ultradian rhythms run shorter than a day, repeating several times within it. A 2016 review in the journal Temperature, "Timing of activities of daily life is jaggy", puts it plainly: in many species "the day is segmented into active periods when the animal searches for food, and inactive periods when the animal digests and rests. This episodic temporal patterning is conventionally referred to as ultradian (<24 hours) rhythmicity."

Two things live under that word. The first is the sleep cycle, which is measured, replicated and taught in every physiology course. The second is the idea that the same cycle keeps running while you are awake, driving alertness and focus. The first is solid. The second is a hypothesis with a famous name attached.

What is solidly established: the night

The sleep cycle is the strongest ultradian rhythm in human biology. The StatPearls physiology chapter on sleep stages, indexed on PubMed, describes two phases, rapid eye movement and non-rapid eye movement sleep, the second divided into three stages, and states: "The body cycles through all stages approximately 4 to 6 times each night, averaging 90 minutes for each cycle."

The National Institute of Neurological Disorders and Stroke says the same in its Brain Basics guide to sleep: "REM sleep first occurs about 90 minutes after falling asleep." It adds the detail the 90-minute slogan leaves out: "You cycle through non-REM and REM sleep several times during a typical night, with increasingly longer, deeper REM periods occurring later in the sleep session." The cycles are not identical. Early ones are heavy in deep sleep, late ones in REM.

That is the established part. Roughly 90 minutes, four to six times, changing shape as the night goes on. Everything you have read about 90-minute work blocks is borrowed from here.

Kleitman's hypothesis: the basic rest-activity cycle

The borrowing has a respectable origin. Nathaniel Kleitman, in whose laboratory REM sleep was discovered, proposed that the sleep cycle was one expression of a cycle running around the clock, which he called the basic rest-activity cycle. He revisited the idea in a 1982 paper in the journal Sleep titled "Basic rest-activity cycle--22 years later".

The hypothesis has held up as a description of animal behaviour, with a caveat that matters for anyone planning a calendar around it. A 2018 chapter in the Handbook of Clinical Neurology, "Thermoregulation and the ultradian basic rest-activity cycle", describes the pattern in rats: "Kleitman, in whose laboratory rapid eye movement sleep was discovered, referred to the temporal pattern as "the basic rest-activity cycle." The active periods of the basic rest-activity cycle occur approximately every 1-2 hours in an irregular stochastic pattern that has been described (perhaps unhelpfully) as ultradian rhythmicity."

Irregular. Stochastic. Perhaps unhelpfully called a rhythm. The 2016 review by the same group is more explicit: "The average time between ultradian events is approximately 1-2 hours, but the interval is highly variable. The ultradian pattern is stochastic, jaggy rather than smooth, so that although the next event is likely to occur within 1-2 hours, it is not possible to predict the precise timing."

A cycle whose next beat cannot be predicted is not a cycle you can schedule meetings around. It is a tendency. A 1999 review of normal human sleep in Respiratory Care Clinics of North America states the broad version: "Human sleep is an integral part of the basic rest-activity cycle that is found in all life forms." That is widely accepted. The leap from there to "your focus peaks every 90 minutes" is not in the literature.

The human daytime evidence: sleepability, not productivity

The best human evidence for a daytime ultradian rhythm comes from a pair of laboratory studies in the 1980s, and it is worth seeing exactly what they measured.

In the first, published in 1981, "9 young adults were placed on an ultrashort 15 min waking-5 min sleeping schedule for 12 consecutive hours." Every twenty minutes they tried to fall asleep for five. The finding: "the amount of stage 1 in each of the sleep attempts varied rhythmically with a frequency of about 14.4 c/day." Fourteen point four cycles a day is one every hundred minutes. The authors interpreted the results "in the light of Kleitman's BRAC model," and both REM deprivation and total sleep deprivation "disrupted the 100 min periodicity in stage 1."

The 1984 follow-up ran eight subjects on the same schedule in the afternoon, evening and morning and found "a dominant ultradian frequency of about 7.2 c/day during both experimental schedules." When the morning data were lined up against each person's last REM period of the night, "a single spectral peak at 14.4 c/day" appeared, "which is the dominant ultradian frequency of the nocturnal REM-non-REM cycles."

Read what was measured. The ability to fall asleep, in a laboratory, under an artificial schedule, in eight or nine young adults. The researchers called it sleepability. Nobody measured output, decision quality or focus. The studies are good evidence that a roughly 100-minute tendency in sleepiness exists in the daytime. They are no evidence at all that your best work arrives in 90-minute packets.

Where the 90-minute work block came from, and what it rests on

Popular writing fused three separate things: the sleep cycle, Kleitman's hypothesis, and the deliberate-practice literature on elite performers. The first is solid. The second is a tendency with unpredictable timing. The third deserves a closer look.

The deliberate-practice research, summarised by its lead author in a 1993 symposium paper, makes a claim about volume over years: "The amount of deliberate practice accumulated during many years of preparation is found to be related closely to performance, even at the highest levels." The widely repeated detail that elite performers work in sessions of about 90 minutes is not in that abstract, and the full texts were not available for this article, so that figure is not established here. Even if it were, a musician's practice session is not a founder's afternoon.

The honest conclusion: block length is a design choice. Biology supports it weakly, with an irregular one-to-two-hour tendency. Attention research supports it strongly, from the other direction, by showing what interruptions cost. A 2008 study of interrupted work, presented at CHI, found that "people completed interrupted tasks in less time with no difference in quality," and then named the price: "Our data suggests that people compensate for interruptions by working faster, but this comes at a price: experiencing more stress, higher frustration, time pressure and effort."

So the case for protected blocks does not need the 90-minute myth. Uninterrupted time is valuable because interruption is expensive, not because a brainstem timer says so. The deep work article builds the full argument.

How Apex blocks the day

This is the Apex protocol, presented as a design choice informed by the evidence above, not as biology.

1. Two or three Build blocks of 60 to 90 minutes, hardest first. Sixty to ninety because the irregular one-to-two-hour tendency is the only biological hint available, and because most operators cannot hold a single hard problem much longer without the quality dropping. The first block goes to the work that moves the business, before the inbox opens.
2. A real break between blocks, 10 to 20 minutes. Real means away from the screen: stand, walk, water, outside if possible. Checking email is not a break; it is a second task, and the 2008 study is about what second tasks cost.
3. One output per block, written before it starts. A block without a defined output becomes browsing with a timer.
4. Shallow work in the dip. The early afternoon is where most people's alertness sags. Calls, admin and messages go there. The brain fog article covers the inputs that make that dip worse than it needs to be.
5. Decide the blocks the evening before. Decision quality drops through the day, so the day's architecture is decided while it is still high. The decision fatigue article covers why.

Training sits inside this structure as the first block of the day, which is why it matters here. A morning session is a 60-minute block with a defined output, a hard stop and a real recovery after it. It teaches the rhythm the rest of the day runs on. The time blocking method and the compound operator's weekly schedule show the full day and week.

Signs you have overrun a block

Since the biology will not send a notification, you need observable signals. These are the ones operators report most often, and each is a cue to stop on purpose rather than drift.

• You read the same paragraph or the same cell range twice.
• You open a second tab, document or chat without deciding to.
• Your posture has collapsed toward the screen and your breathing has gone shallow.
• You are looking for food you do not need.
• The next sentence or decision keeps being postponed by small edits to the last one.

The Stoic response is the same as the operator's: stop deliberately, while the decision is still yours. A block ended on purpose at 70 minutes is a success. A block that dissolves into the inbox at 110 is not.

The one-week experiment to find your block length

The research cannot tell you your number. One week of honest logging can.

1. Day 1 and 2: 60-minute blocks. Two or three per day, defined output, 15-minute real break. At the end of each block, write two numbers: minutes until the first overrun sign appeared, and whether the output was finished (yes or no).
2. Day 3 and 4: 75-minute blocks. Same log.
3. Day 5 and 6: 90-minute blocks. Same log.
4. Day 7: read the log. Your block length is the longest setting at which the overrun sign arrived after the block ended and the output was usually finished. For many operators it is shorter than 90 minutes in the afternoon and longer in the morning. That is allowed. Nothing in the evidence says the number must be constant.

Keep the sleep variable honest while you test. Every source above ties the daytime tendency to the night before, and the 1981 study found that a night of deprivation disrupted the daytime periodicity entirely. A block experiment run on five hours of sleep measures the sleep debt, not the block.

Frequently asked questions

What is an ultradian rhythm?

An ultradian rhythm is a biological cycle shorter than a day that repeats several times within it, as opposed to a circadian rhythm, which runs on roughly 24 hours. The clearest human example is the sleep cycle, which the StatPearls physiology chapter describes as repeating "approximately 4 to 6 times each night, averaging 90 minutes for each cycle."

Is the 90-minute productivity cycle real?

The 90-minute figure is real for sleep cycles. For the waking day, laboratory studies in the 1980s found a roughly 100-minute rhythm in how easily people could fall asleep, in groups of eight or nine young adults. No study in this article measured focus or output on a 90-minute cycle, and reviews describe the daytime pattern as irregular and unpredictable in timing.

What is the basic rest-activity cycle?

The basic rest-activity cycle is Nathaniel Kleitman's hypothesis that the alternation seen in sleep continues through the waking day as cycles of activity and rest. In animals, active periods occur roughly every one to two hours in what a 2018 review calls "an irregular stochastic pattern." It is a well-supported tendency, not a precise clock.

How long should a work block be?

The evidence does not fix a number. Biology offers an irregular one-to-two-hour tendency, and attention studies found that interruptions raise stress and effort. The Apex protocol uses 60 to 90 minute blocks with 10 to 20 minute real breaks, and a one-week experiment at 60, 75 and 90 minutes to find the length that fits you.

Does lack of sleep affect daytime ultradian rhythms?

In the 1981 laboratory study, both REM deprivation and total sleep deprivation "disrupted the 100 min periodicity" in daytime sleepiness. Whatever daytime rhythm you have depends on the night before, which is why any experiment with block length should be run on a normal sleep schedule.

Borrow the discipline, not the myth

The useful part of the ultradian idea survives contact with the evidence. Work in bounded blocks. Rest for real between them. Put the hardest work early. Decide the structure the night before. None of that needs a 90-minute timer in the brainstem to be true, and pretending it does makes the practice brittle the first time your focus lasts 50 minutes instead of 90.

What the evidence does insist on is the night. The one ultradian rhythm that is beyond dispute runs while you sleep, and the daytime tendency, whatever its length, collapses without it. The operator who protects sleep and blocks the day deliberately gets both halves of the system.

The free 5-Day Stoic Operator Challenge installs that structure: a morning training block with a hard stop, a daily review that sets tomorrow's blocks, and a day built on decisions instead of reactions. Run it for five days, then run the block experiment for seven.