Everyone knows sleep follows a roughly 24-hour circadian rhythm. A new study suggests that underneath that daily clock, there's a second, much slower rhythm most people have never had the data to see: sleep quality itself seems to rise and fall in cycles lasting anywhere from about a week to nearly two months.
The findings, now published as "Sleep Homeostasis Has Circaseptan and Circatrigintan Cycles Over Year-Long Timescales" in the Journal of Sleep Research, grew out of the Ultra-Long-Term Sleep study, in which 25 healthy adults (mean age 32, ranging roughly ±13 years) were fitted with a minimally invasive subcutaneous EEG lead — a small electrode implanted just under the scalp that continuously records brain activity without the need for a sleep lab. Twenty participants completed a full year of recording, generating one of the longest continuous sleep datasets ever collected from healthy individuals.
Cycles Within Cycles
Researchers tracked total sleep duration, time spent awake after initially falling asleep, how long it took to fall asleep, and overall sleep efficiency across the entire year for each participant. Rather than fluctuating randomly from night to night, these measures traced out recurring, subject-specific cycles with approximate periodicities clustering around 8 to 11 days, 21 days, 31 days, and 55 to 58 days.
The researchers describe the roughly weekly-scale rhythms as "circaseptan" and the roughly monthly-scale rhythms as "circatrigintan" — borrowing terms long used in other areas of chronobiology, such as hormonal and immune cycles, but rarely applied to sleep itself. Notably, these cycles appeared to be intrinsic to each individual rather than driven by an external schedule like a workweek or a lunar month: they persisted independent of the kinds of factors — weather, travel, changing routines — that might be expected to explain periodic swings in how well someone sleeps.
Separate From the Seasons
The study also found a more familiar pattern layered on top: seasonal shifts in sleep timing. In summer months, participants' sleep onset was delayed by up to 10 minutes and their wake times shifted up to 20 minutes earlier compared with the rest of the year — consistent with longer daylight hours nudging the circadian clock. But the researchers found the multi-day and multi-week cycles were statistically distinct from this seasonal effect, suggesting sleep regulation operates on at least three nested timescales simultaneously: the familiar daily cycle, a seasonal cycle tied to the calendar year, and a previously unrecognized set of intermediate cycles spanning roughly a week to two months.
An Underused Window Into Sleep Biology
Collecting a full year of nightly brain-activity data from healthy people was, until recently, logistically impractical — traditional polysomnography requires a lab, wires, and a technician. The subcutaneous EEG approach used here, developed to let patients wear a credit-card-sized recorder at home, made year-long passive monitoring feasible for the first time in a non-clinical population. That technical shift is itself part of the story: longer-duration, lower-burden recording may reveal structure in human physiology that shorter studies, by design, cannot detect.
What This Means for Patients
For most people, this won't change anything about tonight's sleep routine, but it may eventually reshape how clinicians interpret a bad week or two of sleep. A patient who sleeps worse for several nights in a row, seemingly without an obvious trigger, may not be dealing with a new problem at all — they may simply be in a trough of an intrinsic cycle that will resolve on its own within days to weeks. That has practical implications for how sleep clinics evaluate treatment response: judging whether a new insomnia therapy, sleep medication, or behavioral intervention is working based on a week or two of data could be misleading if an existing multi-week cycle happens to coincide with starting treatment. If these cycles are confirmed and better characterized in future work, accounting for them could make both clinical trials and individual treatment decisions more accurate, and spare patients unnecessary alarm over a rough patch of sleep that was always going to pass.