Nearly every cell you have runs a molecular clock with a period close to 24 hours, and a master timer in the hypothalamus holds those cellular clocks in step by reading light through a small set of retinal cells built for the job. Light is the main signal that sets the system. Morning light pulls the clock earlier and night light pushes it later, which is why the same lamp helps at 7am and hurts at 11pm.
Food is a second signal that sets the clocks in the liver and gut without moving the master clock in the brain. This page describes the machinery; the pages that tell you what to do with it are linked near the end.
Findings & Outcomes
What It Is
Circadian entrainment is the daily correction that locks your internal clock to the outside day. The clock keeps time on its own: left in constant darkness, the master timer in the brain drifts only a few minutes a day. Entrainment is what holds that free-running clock to the 24-hour day, and light is the main signal that does it.
Six other pages in this section act on this one timing system, each on a different input: morning light, evening light and screens, the sleep environment, insomnia, the sleep protocol, and time-restricted eating. This page describes the mechanism they share and recommends nothing; the pages that tell you what to do are linked at the end.
How the Clock Keeps Time
Underneath everything is a loop of gene activity inside a single cell. Two proteins, CLOCK and BMAL1, switch on the Period and Cryptochrome genes. The PER and CRY proteins they make build up through the day, pair off, move back into the nucleus and shut off the very machinery that made them, and are then tagged and broken down so the whole thing starts again. That is the transcription-translation feedback loop, and it runs inside individual cells with no input from the rest of the body: cells taken out and kept in a dish keep oscillating for days.
How a cell keeps a day
1A gene loop that takes about a day to run
CLOCK and BMAL1 turn on the Period and Cryptochrome genes. Their proteins accumulate, then feed back and switch their own production off, then are degraded so the cycle restarts. One turn of that loop takes close to 24 hours, and it is the timekeeping unit that every other layer is built on.
2One amino acid can advance sleep by hours
In familial advanced sleep phase syndrome, an inherited condition, people wake and sleep about four hours early, with their temperature and melatonin rhythms advanced by the same amount. The cause is a single serine-to-glycine change in PER2, in the region a kinase binds, which leaves the protein under-tagged. That one amino acid moves a person's sleep four hours earlier, direct evidence that this loop sets human sleep timing.
3The loop runs in almost every tissue
The same machinery sits in liver, muscle, gut, pancreas, kidney and fat. Across 13 tissues from 632 human donors, nearly half of all protein-coding genes cycle in at least one tissue, found by an algorithm that reconstructs sample order with no time-of-day information. About a thousand of those genes encode drug targets or the proteins that handle drugs.
That method orders samples within a reconstructed cycle instead of against clock time, so it gives phase relationships between genes and between tissues, not an hour of the day for any organ.
The intrinsic period is about 24.2 hours. Almost every popular account gives 25 hours, usually with the claim that the human clock naturally runs longer than the 24-hour day. That figure came from early isolation studies in which participants lived underground for weeks and switched their own room lights on and off, so the light they chose was feeding back onto the clock being measured and lengthening the estimate. Under forced desynchrony, where the schedule sits outside the range the clock can lock onto and light is controlled by the protocol, the intrinsic period averages 24.18 hours, in young and older people alike. Across 157 adults aged 18 to 74 studied in month-long inpatient protocols, the standard deviation was about 12 minutes. The correction was published in 1999. A 25-hour clock would need an hour of correction every day and drift badly the moment light went irregular; a clock running about 12 minutes long needs only 12 minutes of correction, which ordinary daylight supplies without anyone trying.
Peripheral clocks, and food as a second timing signal. In an intact body the master clock keeps the organ clocks aligned through behavior, hormones, the nervous system and the daily swing in body temperature. Food can pull those peripheral clocks out of that alignment. In mice, restricting food to the wrong part of the day shifted clock gene expression in the liver, kidney, heart and pancreas by up to 12 hours while leaving the master clock where it was, and the liver reset fastest. The human effect is smaller: when ten young men delayed every meal by five hours, their plasma glucose rhythm delayed by 5.69 hours and a clock gene in their fat tissue delayed by about an hour, while melatonin and cortisol, the markers of the master clock, held still. Because meals move the organ clocks this way, meal timing is a circadian intervention and the mechanism underneath the time-restricted eating page. The mouse effect is far larger than the human one, and most popular meal-timing claims come from the mouse studies.
Movement is a weaker second cue. Light is the dominant signal, and the clock also responds to the daily rhythm of activity. When 51 older and 48 younger adults did an hour of moderate treadmill exercise on three days at set times, the hour of the workout shifted their melatonin rhythm along a curve of its own: exercise at 7am and again from 1pm to 4pm advanced the clock, and exercise from 7pm to 10pm delayed it. Exercise is a milder timing signal than light and acts alongside it, so the hour of a workout has a modest circadian effect.
Human muscle keeps much of its rhythm from outside signals. Serial muscle biopsies through the day show extensive rhythmic gene activity, and much of that rhythm disappears when cells from the same people are grown in a dish. Part of what looks like a muscle clock is the body's daily signals arriving at the muscle.
Chronotype is the phase your clock settles at once it entrains. It is partly inherited and changes with age. A genome-wide study of 697,828 people raised the number of gene sites linked to morning preference from 24 to 351, though the effect on timing is small: the 5% carrying the most morningness variants slept about 25 minutes earlier than the 5% carrying the fewest. Across 53,689 American time-use diaries, chronotype gets later through adolescence, reaches its latest point at 18.4 years in females and 19.2 in males, then moves earlier for the rest of life. Adolescent lateness is biological, and it reverses on its own.
Sex and circadian biology run in a consistent direction where they have been measured directly. Across the same 157-person dataset, women's intrinsic period averaged 24.09 hours against men's 24.19, a difference of about 6 minutes, and 35% of women had a period shorter than 24.0 hours against 14% of men. In a separate study matching 28 women and 28 men on wake time, women's melatonin and temperature rhythms sat earlier relative to sleep, so identical schedules place men and women at different points of their own biological night. Most of the rest of this field is male-skewed, and the evidence below flags it row by row.
How Light Sets It
The master clock is the suprachiasmatic nucleus, about 10,000 neurons on each side of the hypothalamus, sitting directly above the crossing of the optic nerves. It holds its rhythm in culture, it keeps the peripheral clocks aligned, and it is set by light through a pathway separate from the one you see with.
Alongside rods and cones, the retina carries a small population of intrinsically photosensitive retinal ganglion cells that contain the pigment melanopsin and project straight to the suprachiasmatic nucleus. They respond most strongly to short-wavelength blue light. Two independent action spectra for melatonin suppression converged, and neither fits rod or cone absorption: one found the best fit at a photopigment peaking near 459 nm across 22 volunteers, the other found 446 to 477 nm the most potent region across 72 people. In people, this pathway works separately from sight. Among eleven patients with no conscious perception of light, bright light suppressed melatonin in three, and those three kept normal circadian timing, while the other eight showed no response and reported histories of insomnia and circadian disturbance.
This is also why ordinary lux is the wrong unit for anything to do with the clock. Lux weights light by daytime vision, which peaks in the green. Melanopic lux weights the same light by what melanopsin responds to, and it is the scale on which two lamps that look equally bright can differ severalfold. The applied thresholds sit on the evening light page.
The phase response curve explains why two pages in this section give opposite instructions about the same input. How the clock responds to light depends on when the light arrives, measured against the clock's own phase. The reference point is the core body temperature minimum, which in a regular sleeper falls about two hours before waking.
- Light before the temperature minimum, meaning the evening and first part of the night, delays the clock. Sleep and waking move later.
- Light after the temperature minimum, meaning late in sleep and through the morning, advances the clock. Sleep and waking move earlier.
- The crossover sits at the minimum itself, and the curve is steepest on either side of it and nearly flat through the middle of the day.
The sizes are modest and the shape is non-linear. A bright pulse after the minimum advances the clock along a curve with a peak-to-trough swing of about five hours across the day in one laboratory series, and a one-hour pulse of 8,000 lux produced a swing of 2.20 hours, about 40% of what a 6.7-hour exposure produced on 15% of the exposure time, so the first minutes of light do far more per minute than the last.
Bright light at 11pm and bright light at 7am are opposite interventions.
Someone whose temperature minimum sits at 8am gets a delay from the same 6am light that advances an early riser.
Remove electric light and the clock resets to the natural day. Eight people who spent a week camping in summer with no electrical light saw their biological night shift to begin near sunset and end near sunrise, about two hours earlier than in modern life, with the latest chronotypes advancing most. Repeated in winter, the biological night began earlier still and lasted longer, tracking the shorter photoperiod. Eight people is very few, and this is the cleanest demonstration available that late timing in modern life comes from the light environment, not from the clock itself.
The Research & Studies
Everything here is based on the research we have collected and checked, sorted into groups and ordered with the strongest evidence first. Click any claim to open the studies behind it.
How it works
The human body clock runs about 24.2 hours, close to the length of a day
Measured under forced desynchrony, with the imposed day length outside the range the clock can lock onto and light controlled by the protocol rather than by the participant, the intrinsic period averages 24.18 hours, about 24.2, in young and older adults alike, with a distribution as tight as any other species tested. Across 157 adults studied in month-long inpatient protocols the standard deviation was about 0.2 hours, roughly 12 minutes. The 25-hour figure still in general circulation came from earlier isolation studies in which participants switched their own room lighting, so self-selected light was resetting the clock being measured.
Measured under forced desynchrony, with the imposed day length outside the range the clock can lock onto and light controlled by the protocol rather than by the participant, the intrinsic period averages 24.18 hours, about 24.2, in young and older adults alike, with a distribution as tight as any other species tested. Across 157 adults studied in month-long inpatient protocols the standard deviation was about 0.2 hours, roughly 12 minutes. The 25-hour figure still in general circulation came from earlier isolation studies in which participants switched their own room lighting, so self-selected light was resetting the clock being measured. Measured in: Healthy young and older adults in controlled-lighting inpatient protocols. The pooled period estimate covers 157 people aged 18 to 74, 52 women and 105 men.. Forced desynchrony measures the pacemaker's own period by imposing a schedule nobody lives on, and it costs weeks of inpatient time per person, which is why the total number of people ever measured this way is in the low hundreds. It gives the period of the clock, not the timing anyone actually keeps, which is set by light on top of that period.
Who this may not transfer to:Two thirds of the pooled sample were men (105 to 52), and the sex difference in period is reported separately in the row below rather than folded into this average.
The studies · 2
Czeisler et al., stability, precision, and near-24-hour period of the human circadian pacemaker · Science 1999;284(5423):2177-81
Duffy et al., sex difference in the near-24-hour intrinsic period of the human circadian timing system · Proc Natl Acad Sci USA 2011;108 Suppl 3:15602-8
Counts once: this finding and 1 other here come from the same source, so they are one body of evidence, not separate confirmations.
A retinal pigment, melanopsin, carries blue light to the clock along a pathway separate from vision
Two independent action spectra for nocturnal melatonin suppression fit a photopigment distinct from rods and cones. One, from 22 volunteers across 215 monochromatic light exposures, fit best to a rhodopsin template peaking at 459 nm, with rod, cone and cryptochrome spectra all fitting poorly. The other, from 72 people across 627 suppression tests, found 446 to 477 nm the most potent region and fit an opsin template with an R-squared of 0.91. The pigment is melanopsin, carried by a small population of intrinsically photosensitive retinal ganglion cells that project to the suprachiasmatic nucleus.
Two independent action spectra for nocturnal melatonin suppression fit a photopigment distinct from rods and cones. One, from 22 volunteers across 215 monochromatic light exposures, fit best to a rhodopsin template peaking at 459 nm, with rod, cone and cryptochrome spectra all fitting poorly. The other, from 72 people across 627 suppression tests, found 446 to 477 nm the most potent region and fit an opsin template with an R-squared of 0.91. The pigment is melanopsin, carried by a small population of intrinsically photosensitive retinal ganglion cells that project to the suprachiasmatic nucleus. Measured in: 22 volunteers over 215 light exposure trials, and 72 volunteers (37 women, 35 men, mean age 24.5) over 627 melatonin suppression tests. Action spectra are built from melatonin suppression at night, which is one output of the pathway rather than the pathway itself, and both used narrow-band light at intensities and durations nobody encounters. Melanopsin's own peak sensitivity in the intact eye sits nearer 480 nm than these melatonin figures, because rods and cones also feed into the same ganglion cells.
Who this may not transfer to:The larger study was 37 women and 35 men; the smaller one does not report its sex composition. Everyone measured was a young adult, and lens transmission of short-wavelength light falls steadily with age, so the same spectrum delivers less to the clock in an older eye.
The studies · 2
Thapan, Arendt and Skene, an action spectrum for melatonin suppression · J Physiol 2001;535(Pt 1):261-7
Brainard et al., action spectrum for melatonin regulation in humans · J Neurosci 2001;21(16):6405-12
Each cell keeps time with a gene feedback loop that takes about a day to run
The clock is a transcription-translation feedback loop. CLOCK and BMAL1 bind and drive transcription of the Period and Cryptochrome genes; the PER and CRY proteins accumulate, complex, return to the nucleus and repress the very activators that made them, and are then phosphorylated by kinases and degraded so the cycle restarts. The loop is cell-autonomous and keeps running in isolated cells. In humans, a single serine-to-glycine substitution in PER2, inside the casein kinase I epsilon binding region, produces autosomal dominant familial advanced sleep phase syndrome, with a four-hour advance of the sleep, temperature and melatonin rhythms.
The clock is a transcription-translation feedback loop. CLOCK and BMAL1 bind and drive transcription of the Period and Cryptochrome genes; the PER and CRY proteins accumulate, complex, return to the nucleus and repress the very activators that made them, and are then phosphorylated by kinases and degraded so the cycle restarts. The loop is cell-autonomous and keeps running in isolated cells. In humans, a single serine-to-glycine substitution in PER2, inside the casein kinase I epsilon binding region, produces autosomal dominant familial advanced sleep phase syndrome, with a four-hour advance of the sleep, temperature and melatonin rhythms. Measured in: Molecular work in flies, mice and cultured cells; the human genetic evidence comes from a kindred with familial advanced sleep phase syndrome. The loop was worked out mostly outside humans, and the human genetic evidence rests on a small number of families carrying rare high-effect mutations. That shows the loop sets human sleep timing. It does not show how much of ordinary chronotype variation runs through these same genes, and the genome-wide work suggests common variation moves timing by minutes rather than hours.
Who this may not transfer to:The PER2 kindred includes affected men and women, as an autosomal dominant trait would predict. The molecular work underneath has no participant sex basis, being done in flies, mice and cultured cells.
The studies · 2
Takahashi, transcriptional architecture of the mammalian circadian clock · Nat Rev Genet 2017;18(3):164-79
Toh et al., an hPer2 phosphorylation site mutation in familial advanced sleep phase syndrome · Science 2001;291(5506):1040-3
Women's body clocks run about 6 minutes shorter than men's and sit at an earlier phase
Women's intrinsic period averaged 24.09 hours against men's 24.19 hours (P < 0.01), a difference of about 6 minutes, and 35% of women had a period shorter than 24.0 hours against 14% of men. In a separate constant-routine study of 28 women and 28 men matched on habitual wake time, the melatonin and core body temperature rhythms occurred earlier relative to sleep in women, with higher melatonin amplitude and lower temperature amplitude, so women were sleeping at a later biological phase on the same clock time.
Women's intrinsic period averaged 24.09 hours against men's 24.19 hours (P < 0.01), a difference of about 6 minutes, and 35% of women had a period shorter than 24.0 hours against 14% of men. In a separate constant-routine study of 28 women and 28 men matched on habitual wake time, the melatonin and core body temperature rhythms occurred earlier relative to sleep in women, with higher melatonin amplitude and lower temperature amplitude, so women were sleeping at a later biological phase on the same clock time. Measured in: 157 adults aged 18 to 74 (52 women, 105 men) for the period comparison; 56 adults aged 18 to 30 (28 women, 28 men) for the phase angle comparison. Six minutes of period difference and a phase difference well under an hour: replicated across two datasets, and small next to what an alarm clock imposes. Both datasets come from the same research group and overlapping laboratory protocols, so these are related analyzes rather than independent confirmation. Menstrual cycle phase and hormonal contraception were not controlled across the whole period sample.
Who this may not transfer to:This is one of the few rows on this page where both sexes were measured and compared directly, which is the reason it exists as its own claim rather than as a note under another.
The studies · 2
Duffy et al., sex difference in the near-24-hour intrinsic period of the human circadian timing system · Proc Natl Acad Sci USA 2011;108 Suppl 3:15602-8
Cain et al., sex differences in phase angle of entrainment and melatonin amplitude in humans · J Biol Rhythms 2010;25(4):288-96
Counts once: this finding and 1 other here come from the same source, so they are one body of evidence, not separate confirmations.
Light after the temperature minimum advances the clock, light before it delays the clock
Light given after the core body temperature minimum advances the clock, and a bright pulse produces a phase-response curve with a peak-to-trough amplitude of about 5 hours under laboratory conditions, and light given before it delays the clock. The crossover sits at the temperature minimum, which falls about two hours before habitual waking in a regular sleeper, and the curve is nearly flat through the middle of the biological day. Response is steeply non-linear with duration: a one-hour pulse of about 8,000 lux produced a curve with a peak-to-trough amplitude of 2.20 hours, roughly 40% of what a 6.7-hour exposure produced on 15% of the exposure time.
Light given after the core body temperature minimum advances the clock, and a bright pulse produces a phase-response curve with a peak-to-trough amplitude of about 5 hours under laboratory conditions, and light given before it delays the clock. The crossover sits at the temperature minimum, which falls about two hours before habitual waking in a regular sleeper, and the curve is nearly flat through the middle of the biological day. Response is steeply non-linear with duration: a one-hour pulse of about 8,000 lux produced a curve with a peak-to-trough amplitude of 2.20 hours, roughly 40% of what a 6.7-hour exposure produced on 15% of the exposure time. Measured in: 21 adults aged 19 to 44 (16 men, 7 women) for the three-cycle curve; 36 adults randomized to bright white or dim light for the one-hour curve. The human phase response literature is small, young and laboratory-bound, and both curves were built from single pulses against a dim controlled background rather than from the mixed light of an ordinary day. It measures the clock moving. The step from a phase shift to sleeping better is an inference from mechanism, not something these protocols measured.
Who this may not transfer to:The larger of the two curves ran 16 men to 7 women, and the one-hour study does not report its sex split by condition. Circadian phase and light sensitivity both change with age and across the menstrual cycle, so older readers and women are thinly represented in the shape of this curve.
The studies · 2
Khalsa et al., a phase response curve to single bright light pulses in human subjects · J Physiol 2003;549(Pt 3):945-52
St Hilaire et al., human phase response curve to a 1 h pulse of bright white light · J Physiol 2012;590(13):3035-45
Some people with no conscious sight still set their clock by light
Among 11 patients with no conscious perception of light, plasma melatonin fell during bright light exposure in three, and those three were normally entrained and slept normally. Seven showed no melatonin response to light and reported histories of insomnia and circadian rhythm disturbance; one had undetectable melatonin. Six sighted controls suppressed normally.
Among 11 patients with no conscious perception of light, plasma melatonin fell during bright light exposure in three, and those three were normally entrained and slept normally. Seven showed no melatonin response to light and reported histories of insomnia and circadian rhythm disturbance; one had undetectable melatonin. Six sighted controls suppressed normally. Measured in: 11 blind patients with no conscious light perception and 6 sighted controls. Eleven patients with different causes and degrees of blindness, so this establishes that photic input to the clock and conscious vision can come apart, and it gives no reliable proportion for how often. It also predates the identification of melanopsin by seven years, so the mechanism was inferred here rather than measured.
Who this may not transfer to:The sex composition of the blind cohort is not recoverable from the record we checked, so whether the finding rests on a balanced sample is unknown rather than shown to be balanced.
The study · 1
Czeisler et al., suppression of melatonin secretion in some blind patients by exposure to bright light · N Engl J Med 1995;332(1):6-11
Nearly half of protein-coding genes cycle daily in at least one tissue
Applying an algorithm that reconstructs the temporal order of samples with no time-of-day information available, across 13 tissues from 632 human donors, nearly half of all protein-coding genes were found cycling in at least one tissue. About 1,000 of those encode drug targets, or the proteins that transport and metabolize drugs.
Applying an algorithm that reconstructs the temporal order of samples with no time-of-day information available, across 13 tissues from 632 human donors, nearly half of all protein-coding genes were found cycling in at least one tissue. About 1,000 of those encode drug targets, or the proteins that transport and metabolize drugs. Measured in: 13 tissues from 632 human donors in a post-mortem genotype-expression resource. The method orders samples within a reconstructed cycle rather than against clock time, so it produces phase relationships between genes and between tissues and not an hour of the day for any organ. Donor tissue also varies in agonal state, cause of death and time of collection, and the reconstruction is an inference from expression rather than a rhythm watched in a living person.
Who this may not transfer to:The donor sex composition is not reported in the abstract, and the underlying tissue resource is male-weighted, so tissue rhythms specific to one sex would not be visible in this summary.
The study · 1
Ruben et al., a database of tissue-specific rhythmically expressed human genes · Sci Transl Med 2018;10(458):eaat8806
In mice, daytime feeding shifts organ clocks up to 12 hours while the brain clock holds
Restricting food to the light phase shifted the phase of clock gene expression in mouse peripheral tissues by up to 12 hours while leaving cyclic gene expression in the suprachiasmatic nucleus unaffected. The liver reset fastest, with kidney, heart and pancreas following more slowly, and one week of daytime feeding synchronized every tissue examined.
Restricting food to the light phase shifted the phase of clock gene expression in mouse peripheral tissues by up to 12 hours while leaving cyclic gene expression in the suprachiasmatic nucleus unaffected. The liver reset fastest, with kidney, heart and pancreas following more slowly, and one week of daytime feeding synchronized every tissue examined. Measured in: Mice on restricted feeding schedules, with clock gene expression measured in liver, kidney, heart, pancreas and the suprachiasmatic nucleus. Mice are nocturnal, so feeding them by day is a larger insult than late eating is to a person, and the human version of the same manipulation produced a peripheral shift a fraction of this size. Animal work establishes that the route exists rather than what it does in people.
The study · 1
Damiola et al., restricted feeding uncouples circadian oscillators in peripheral tissues from the central pacemaker in the suprachiasmatic nucleus · Genes Dev 2000;14(23):2950-61
Living 12 hours out of phase raised glucose 6% and insulin 22% in the laboratory
On a 28-hour day protocol that walked eating and sleeping across every circadian phase, behavior running roughly 12 hours out of phase with the internal clock raised glucose 6% (P < 0.001), insulin 22% (P = 0.006) and mean arterial pressure 3% (P = 0.001), lowered leptin 17% (P < 0.001) and completely reversed the daily cortisol rhythm. Three of eight participants showed postprandial glucose in the prediabetic range while misaligned. Sleep efficiency fell 20%.
On a 28-hour day protocol that walked eating and sleeping across every circadian phase, behavior running roughly 12 hours out of phase with the internal clock raised glucose 6% (P < 0.001), insulin 22% (P = 0.006) and mean arterial pressure 3% (P = 0.001), lowered leptin 17% (P < 0.001) and completely reversed the daily cortisol rhythm. Three of eight participants showed postprandial glucose in the prediabetic range while misaligned. Sleep efficiency fell 20%. Measured in: 10 adults, 5 women and 5 men, through a 10-day inpatient forced desynchrony protocol. Ten people over ten days, with food, activity and sleep opportunity held constant, which isolates misalignment itself and says nothing about what years of it do. Sleep efficiency also fell 20%, so disturbed sleep sits inside the misaligned condition rather than beside it, and the two cannot be separated in this design.
Who this may not transfer to:Five women and five men, which is an even split and far too small to compare them. Results are not reported split by sex.
The study · 1
Scheer et al., adverse metabolic and cardiovascular consequences of circadian misalignment · Proc Natl Acad Sci USA 2009;106(11):4453-8
Delaying meals 5 hours shifted the glucose rhythm nearly 6 hours without moving the master clock
Delaying every meal by five hours delayed the plasma glucose rhythm by 5.69 hours (P < 0.001) and the PER2 messenger RNA rhythm in adipose tissue by 0.97 hours (P < 0.01), while plasma melatonin and cortisol, the markers of the master clock, did not shift at all. Average glucose concentration fell 4.9 mg/dL (0.27 mmol/L). Insulin and triglyceride rhythms, whole-blood clock gene expression, sleep and the rhythms of subjective hunger and sleepiness did not move.
Delaying every meal by five hours delayed the plasma glucose rhythm by 5.69 hours (P < 0.001) and the PER2 messenger RNA rhythm in adipose tissue by 0.97 hours (P < 0.01), while plasma melatonin and cortisol, the markers of the master clock, did not shift at all. Average glucose concentration fell 4.9 mg/dL (0.27 mmol/L). Insulin and triglyceride rhythms, whole-blood clock gene expression, sleep and the rhythms of subjective hunger and sleepiness did not move. Measured in: 10 healthy young men, mean age 22.9, in a crossover with early meals and then meals delayed by five hours. Ten men, one laboratory, and the two schedules ran in a fixed order rather than randomized, so an order effect cannot be excluded. The adipose clock gene moved by about an hour against a five-hour meal delay, a far smaller peripheral shift than the same manipulation produces in rodents, and adipose tissue is a proxy for the liver and gut clocks that cannot be biopsied as easily.
Who this may not transfer to:All ten participants were men. Circadian period, phase and the glucose response to a meal all differ on average between the sexes, and this design could not have detected that, so a reader who is not a man should treat the size of these shifts as unmeasured for her rather than established.
The study · 1
Wehrens et al., meal timing regulates the human circadian system · Curr Biol 2017;27(12):1768-75
Much of the daily rhythm in human muscle comes from the body's signals, not the muscle's own clock
Serial muscle biopsies taken through the day in humans showed extensive rhythmic transcription, and a large part of that rhythmicity was lost when primary myotubes from the same source were synchronized and studied in culture. Disrupting the clock in those cells with siRNA changed the expression of about 8% of all genes.
Serial muscle biopsies taken through the day in humans showed extensive rhythmic transcription, and a large part of that rhythmicity was lost when primary myotubes from the same source were synchronized and studied in culture. Disrupting the clock in those cells with siRNA changed the expression of about 8% of all genes. Measured in: Human volunteers undergoing serial in vivo muscle biopsies, compared with primary human myotubes synchronized in vitro. The comparison is between muscle inside a living body and cells in a dish, which differ in far more than their clocks, so the rhythmicity missing in vitro shows that systemic daily signals contribute without quantifying how much. Participant numbers and sex composition are not recoverable from the abstract we checked.
Who this may not transfer to:The sex composition of the biopsy donors is not stated in the abstract, so whether muscle rhythms differ between the sexes is unmeasured here rather than shown to be similar.
The study · 1
Perrin et al., transcriptomic analyzes reveal rhythmic and CLOCK-driven pathways in human skeletal muscle · eLife 2018;7:e34114
Timed exercise shifts the clock, advancing it in the morning and afternoon and delaying it in the evening
Exercise, not only light, can nudge the body clock, and the time of day it is done sets which way. When adults exercised for an hour at set times, morning and early-afternoon exercise moved their clock earlier and evening exercise moved it later. Exercise is a weaker timing signal than light and works best together with it, so it is a useful add-on for adjusting to travel or a shifted schedule rather than a stand-in for morning daylight.
Light is the dominant zeitgeber, and the pacemaker also responds to the daily rhythm of behavior. Exercise raises core body temperature, heart rate and a range of hormones, and the timing of that daily signal feeds back onto the clock. The measured exercise phase-response curve advances the clock after the morning and again in the early afternoon and delays it in the evening, which is why the time of a workout, not only whether one happens, carries a circadian effect.
Who this may not transfer to:Both sexes were studied and compared directly, about 55 women and 44 men across the 99 participants, and the exercise phase-response curve showed no significant sex difference in amplitude or waveform.
The study · 1
Youngstedt, Elliott and Kripke, human circadian phase-response curves for exercise · J Physiol 2019;597(8):2253-68
Cancer Risk And Outcome
In 1.4 million women, night shift work showed no effect on breast cancer risk (RR 0.99)
Pooling three prospective UK cohorts with seven previously published prospective studies, covering about 1.4 million women and 4,660 breast cancers among shift workers, the relative risk of breast cancer for any night shift work was 0.99 (95% CI 0.95 to 1.03). Long-term night shift work showed no excess either. The authors concluded that night shift work, including long-term shift work, has little or no effect on breast cancer incidence.
Pooling three prospective UK cohorts with seven previously published prospective studies, covering about 1.4 million women and 4,660 breast cancers among shift workers, the relative risk of breast cancer for any night shift work was 0.99 (95% CI 0.95 to 1.03). Long-term night shift work showed no excess either. The authors concluded that night shift work, including long-term shift work, has little or no effect on breast cancer incidence. Measured in: About 1.4 million women across 10 prospective studies, with night shift work reported by questionnaire. Night shift work was captured by questionnaire, mostly at a single point in time, so exposure misclassification would push a real association towards the null. It tests breast cancer only: the prostate and colorectal associations IARC cites are not addressed here, and neither is the animal or mechanistic evidence that carries much of the classification.
Who this may not transfer to:Breast cancer in women. Male breast cancer, prostate cancer and colorectal cancer are not addressed, so this null result does not carry to the classification as a whole or to men.
The study · 1
Travis et al., night shift work and breast cancer incidence, three prospective studies and meta-analysis · J Natl Cancer Inst 2016;108(12):djw169
IARC calls night shift work probably carcinogenic, Group 2A, on limited human evidence
The IARC working group that met in June 2019 classified night shift work in Group 2A, probably carcinogenic to humans. The classification rests on three separate evidence streams: limited evidence in humans, with positive associations for cancers of the breast, prostate, colon and rectum; sufficient evidence in experimental animals for the carcinogenicity of alteration of the light-dark schedule; and strong mechanistic evidence in experimental systems, based on effects consistent with immunosuppression, chronic inflammation and cell proliferation.
The IARC working group that met in June 2019 classified night shift work in Group 2A, probably carcinogenic to humans. The classification rests on three separate evidence streams: limited evidence in humans, with positive associations for cancers of the breast, prostate, colon and rectum; sufficient evidence in experimental animals for the carcinogenicity of alteration of the light-dark schedule; and strong mechanistic evidence in experimental systems, based on effects consistent with immunosuppression, chronic inflammation and cell proliferation. Measured in: A working group evaluating the published human, animal and mechanistic literature on night shift work, rather than a participant sample. In IARC's own vocabulary, limited evidence in humans means a positive association was observed but chance, bias and confounding could not be ruled out as explanations. A Group 2A classification is a hazard identification: it says an agent can cause cancer under some conditions and carries no estimate of how much risk any particular job or schedule involves. Much of the weight sits on animal and mechanistic work rather than on the human studies.
Who this may not transfer to:The human evidence spans sites specific to each sex, breast and prostate, alongside colorectal cancer in both. The unit evaluated is a published literature rather than a participant group, so the underlying sex balance varies by study and by cancer site.
The studies · 2
IARC Monographs Volume 124, night shift work · IARC Monographs on the Identification of Carcinogenic Hazards to Humans, Vol 124, Lyon 2020
IARC Monographs Vol 124 group, carcinogenicity of night shift work · Lancet Oncol 2019;20(8):1058-9
Sleep
A week under natural light with no electric light moved the biological night about 2 hours earlier
A week camping in summer with no electrical light raised light exposure to about four times the usual and moved the internal biological night roughly two hours earlier, so that it began near sunset and ended near sunrise. Late chronotypes advanced most, ending the week closer to the early types. Repeated under a natural winter light-dark cycle, the biological night and sleep began earlier still and the biological night expanded. A weekend of natural light achieved about 69% of the shift a full week produced.
A week camping in summer with no electrical light raised light exposure to about four times the usual and moved the internal biological night roughly two hours earlier, so that it began near sunset and ended near sunrise. Late chronotypes advanced most, ending the week closer to the early types. Repeated under a natural winter light-dark cycle, the biological night and sleep began earlier still and the biological night expanded. A weekend of natural light achieved about 69% of the shift a full week produced. Measured in: 8 adults, 6 men and 2 women, mean age 30, in the summer study, with a separate winter and weekend study by the same group. Eight people in the summer study, and camping removes electrical light along with work, commuting, screens, indoor temperature and the whole daily routine, so the light change is confounded with everything else that changes on a camping trip. The two studies come from one research group, so they are related work rather than independent replication.
Who this may not transfer to:Six of the eight summer participants were men, and the study was far too small to compare the sexes. The winter and weekend study does not report its composition in the abstract.
The studies · 2
Wright et al., entrainment of the human circadian clock to the natural light-dark cycle · Curr Biol 2013;23(16):1554-8
Stothard et al., circadian entrainment to the natural light-dark cycle across seasons and the weekend · Curr Biol 2017;27(4):508-13
351 gene sites shift sleep timing, but only about 25 minutes between the extremes
A genome-wide association study of 697,828 people raised the number of loci associated with being a morning person from 24 to 351. The effect on timing is smaller than that number suggests: the 5% of participants carrying the most morningness alleles had accelerometer-measured sleep timing about 25 minutes earlier than the 5% carrying the fewest. Associated genes were enriched for circadian regulation, cAMP, glutamate and insulin signaling, and for expression in retina, hindbrain, hypothalamus and pituitary. Mendelian randomization supported a causal path from morning preference to better mental health, and found no effect on BMI or type 2 diabetes.
A genome-wide association study of 697,828 people raised the number of loci associated with being a morning person from 24 to 351. The effect on timing is smaller than that number suggests: the 5% of participants carrying the most morningness alleles had accelerometer-measured sleep timing about 25 minutes earlier than the 5% carrying the fewest. Associated genes were enriched for circadian regulation, cAMP, glutamate and insulin signaling, and for expression in retina, hindbrain, hypothalamus and pituitary. Mendelian randomization supported a causal path from morning preference to better mental health, and found no effect on BMI or type 2 diabetes. Measured in: 697,828 adults from UK Biobank and 23andMe, with activity-monitor sleep timing in 85,760 of them. What could explain it instead: Chronotype was self-reported by most participants, so the phenotype is partly a report of the schedule a person keeps, which work and family impose, rather than the clock underneath it. UK Biobank and 23andMe participants are also healthier, wealthier and more European in ancestry than the general population, the standard population-stratification problem in large genetic studies.. 351 loci that together move sleep timing by about 25 minutes between the extreme twentieths, so common genetic variation shifts chronotype by minutes for most people rather than by hours. Mendelian randomization assumes the genetic instrument affects the outcome only through chronotype, which cannot be verified directly, and the mental health outcomes were themselves self-reported.
Who this may not transfer to:Both sexes are represented in large numbers, and the headline results are not reported split by sex, so a sex difference in the genetic architecture of chronotype would not be visible here.
The study · 1
Jones et al., genome-wide association analyzes of chronotype in 697,828 individuals · Nat Commun 2019;10(1):343
Chronotype peaks latest at about 19 years, then moves earlier for the rest of life
Across 53,689 American time-use diaries, chronotype gets later through adolescence, reaches its latest point at 18.4 years in females and 19.2 years in males, and moves earlier from then on for the rest of life. The largest sex difference is 0.27 hours, about 16 minutes, at 21.6 years, with men later than women before about 40 and women later after it. The difference disappears at 41.4 years and again at 79.1. Chronotype variability also narrows with age.
Across 53,689 American time-use diaries, chronotype gets later through adolescence, reaches its latest point at 18.4 years in females and 19.2 years in males, and moves earlier from then on for the rest of life. The largest sex difference is 0.27 hours, about 16 minutes, at 21.6 years, with men later than women before about 40 and women later after it. The difference disappears at 41.4 years and again at 79.1. Chronotype variability also narrows with age. Measured in: 53,689 respondents to the American Time Use Survey, pooled across 2003 to 2014. What could explain it instead: Mid-sleep on free days from a time-use diary is a behavioral proxy for chronotype, and free-day sleep is still shaped by work, caring duties, school schedules and shift patterns, all of which change with age, so part of the age curve is life stage rather than clock.. A snapshot of different people at different ages rather than the same people followed, so an age effect cannot be separated from a cohort effect. The original European description of the adolescent peak, in about 25,000 questionnaires, placed maximum lateness nearer 20 than 19, so the exact turning point moves with the sample and the instrument.
Who this may not transfer to:The analysis is reported separately for each sex across the whole age range, which is unusual in this literature and is why the sex-by-age crossover can be stated at all.
The studies · 2
Fischer et al., chronotypes in the US, influence of age and sex · PLoS One 2017;12(6):e0178782
Roenneberg et al., a marker for the end of adolescence · Curr Biol 2004;14(24):R1038-9
Morning bright light shifted sleep earlier and eased sleep disturbance across 40 studies
Getting bright light into your eyes in the morning helps you fall asleep earlier and sleep more soundly. Across 40 studies, morning light shifted sleep timing so people drifted off sooner, cut broken sleep, and eased self-rated sleep problems. The gains per study were modest, and keeping the evening dim helped people sleep longer. Outdoor daylight is free; a bright lamp works when you are indoors or on nights.
Light is the master signal that sets the suprachiasmatic clock in the brain. Bright light in the morning, sensed by melanopsin cells in the retina, signals daytime to the clock and shifts its timing earlier (a phase advance), which pulls the natural sleep window forward so sleep onset comes sooner. Dim evenings preserve the melatonin rise that opens that window, which is why evening light avoidance added to the benefit.
Who this may not transfer to:Pooled trials included men and women across a range of ages and diagnoses; sex-specific effects were not separately reported.
The study · 1
Faulkner SM, Bee PE, Meyer N, Dijk DJ, Drake RJ. Light therapies to improve sleep in intrinsic circadian rhythm sleep disorders and neuro-psychiatric illness: A systematic review and meta-analysis. · Sleep Med Rev 2019;46:108-123
Longevity And Mortality
Evening types show about 10% higher death rates, which fades among non-smokers who drink little
In UK Biobank, 433,268 adults followed a mean 6.5 years with 10,534 deaths, definite evening types had a hazard ratio for all-cause mortality of 1.10 (95% CI 1.02 to 1.18) against definite morning types, after adjustment for age, sex, ethnicity, smoking, BMI, sleep duration, socioeconomic status and comorbidity. In the Finnish Twin Cohort, 23,854 people followed 37 years with 8,728 deaths, the adjusted hazard ratio was 1.09 (1.01 to 1.18) and it attenuated mainly through smoking and alcohol, with no excess mortality among non-smokers who were at most light drinkers. Those authors concluded there is little or no independent contribution of chronotype to mortality.
In UK Biobank, 433,268 adults followed a mean 6.5 years with 10,534 deaths, definite evening types had a hazard ratio for all-cause mortality of 1.10 (95% CI 1.02 to 1.18) against definite morning types, after adjustment for age, sex, ethnicity, smoking, BMI, sleep duration, socioeconomic status and comorbidity. In the Finnish Twin Cohort, 23,854 people followed 37 years with 8,728 deaths, the adjusted hazard ratio was 1.09 (1.01 to 1.18) and it attenuated mainly through smoking and alcohol, with no excess mortality among non-smokers who were at most light drinkers. Those authors concluded there is little or no independent contribution of chronotype to mortality. Measured in: 433,268 UK adults aged 38 to 73 and 23,854 Finnish adults, both sexes, followed 6.5 and 37 years respectively. What could explain it instead: Evening preference travels with smoking, heavier drinking, shorter sleep, shift work and lower socioeconomic position. The Finnish stratified analysis is the informative one here: the association disappeared among people who did not smoke and drank little, rather than merely shrinking, which is what residual confounding by those behaviors looks like.. Chronotype was a single self-reported item on a four-point scale, collected once, in both cohorts. A hazard ratio near 1.10 is small enough that unmeasured differences in behavior are a sufficient explanation, and the two cohorts differ in follow-up length by a factor of five, so they are not measuring the same exposure window.
Who this may not transfer to:Both cohorts include men and women in large numbers. Neither headline hazard ratio is reported split by sex, so whether the smoking and alcohol explanation holds equally in both is untested.
The studies · 2
Knutson and von Schantz, associations between chronotype, morbidity and mortality in the UK Biobank cohort · Chronobiol Int 2018;35(8):1045-53
Hublin and Kaprio, chronotype and mortality, a 37-year follow-up study in Finnish adults · Chronobiol Int 2023;40(7):841-849
Heart And Vascular
Shift work tracks with about 23% higher heart attack risk and modestly higher diabetes
Across 34 studies and 2,011,935 people, shift work was associated with myocardial infarction (RR 1.23, 95% CI 1.15 to 1.31), coronary events (RR 1.24, 1.10 to 1.39) and ischemic stroke (RR 1.05, 1.01 to 1.09), with no association with all-cause mortality. Across 12 studies and 226,652 people, shift work was associated with diabetes at a pooled odds ratio of 1.09 (1.05 to 1.12), higher in men (1.37, 1.20 to 1.56) than in women (1.09, 1.04 to 1.14) and higher for rotating shifts.
Across 34 studies and 2,011,935 people, shift work was associated with myocardial infarction (RR 1.23, 95% CI 1.15 to 1.31), coronary events (RR 1.24, 1.10 to 1.39) and ischemic stroke (RR 1.05, 1.01 to 1.09), with no association with all-cause mortality. Across 12 studies and 226,652 people, shift work was associated with diabetes at a pooled odds ratio of 1.09 (1.05 to 1.12), higher in men (1.37, 1.20 to 1.56) than in women (1.09, 1.04 to 1.14) and higher for rotating shifts. Measured in: 2,011,935 people across 34 studies for vascular events, and 226,652 people with 14,595 diabetes cases across 12 studies for diabetes. Both pool observational studies, in which shift work travels with lower socioeconomic position, more smoking, shorter and more fragmented sleep, irregular eating and less daylight, none of which the pooled analyzes can separate from the clock. The relative risks are modest, and the absence of any association with mortality sits oddly beside a raised risk of myocardial infarction.
Who this may not transfer to:The diabetes analysis reports its result separately by sex and finds the association substantially larger in men, which the authors attribute in part to different shift patterns and occupations. The vascular analysis is not broken down by sex.
The studies · 2
Vyas et al., shift work and vascular events, systematic review and meta-analysis · BMJ 2012;345:e4800
Gan et al., shift work and diabetes mellitus, a meta-analysis of observational studies · Occup Environ Med 2015;72(1):72-8
Mood & stress
Bright light therapy lowered depression scores in non-seasonal depression (SMD -0.62)
Bright light in the morning lifts mood, and not only in winter depression. Pooling nine trials, light therapy meaningfully lowered depression scores, with the strongest results from treatment lasting two to five weeks. It helped whether used on its own or alongside other treatment. The effect was mild to moderate, so it works as one steady support rather than a stand-alone cure.
Mood tracks the body clock. Depression often comes with a flattened or misaligned circadian rhythm and disrupted sleep, and timed morning light appears to work by re-anchoring that rhythm and its downstream signals, including melatonin timing and serotonergic tone. Morning light advances the clock most strongly, which is the usual reason morning timing is recommended, though the pooled trials did not settle the best time of day.
Who this may not transfer to:Trials enrolled men and women; effects were not reported separately by sex, and the perinatal (women-only) subgroup did not show benefit.
The study · 1
Al-Karawi D, Jubair L. Bright light therapy for nonseasonal depression: Meta-analysis of clinical trials. · J Affect Disord 2016;198:64-71
Weight And Fat Loss
Social jetlag tracks with higher body mass index beyond sleep duration, most in the overweight and obese
In a large chronotype database, social jetlag, defined as the difference between mid-sleep on work days and mid-sleep on free days, was associated with higher body mass index over and above sleep duration. The association was carried by people who were already overweight or obese.
In a large chronotype database, social jetlag, defined as the difference between mid-sleep on work days and mid-sleep on free days, was associated with higher body mass index over and above sleep duration. The association was carried by people who were already overweight or obese. Measured in: Respondents to the Munich ChronoType Questionnaire, a large self-selected online sample of both sexes. What could explain it instead: Social jetlag is largest in people whose work starts early relative to their own chronotype, which tracks shift work, manual occupations and lower socioeconomic position, each independently associated with higher BMI. Reverse causation is also open, since sleeping in on free days is a response to accumulated sleep debt, which higher body weight and sleep-disordered breathing both worsen.. Cross-sectional, with self-reported sleep times and self-reported height and weight, so it shows an association rather than cause. The full text is paywalled, so only the abstract-level finding can be verified here: social jetlag is associated with higher BMI beyond the effect of sleep duration, especially in the overweight and obese.
Who this may not transfer to:A large self-selected online sample of both sexes. The BMI association is not reported split by sex, and menstrual cycle, pregnancy and menopause all affect both sleep timing and weight without being modeled.
The study · 1
Roenneberg et al., social jetlag and obesity · Curr Biol 2012;22(10):939-43
Blood Sugar
Eating in an early daytime window improved insulin sensitivity and blood pressure without weight loss
Eating most of your food earlier in the day, in step with daylight, helps your metabolism. When men with prediabetes ate all their meals in a 6-hour morning-to-afternoon window, their insulin sensitivity and blood pressure improved in 5 weeks even though their weight stayed the same. In a larger trial, an early eating window plus lighter portions took off about 5 lb (2.3 kg) more than dieting alone. Front-loading meals is free and needs no special food.
The body's insulin response, and the clocks inside muscle and liver, are tuned for daytime fuel. Glucose tolerance is naturally higher in the morning and falls in the evening, so calories eaten early are handled more efficiently than the same calories eaten late. Aligning the eating window with the light-day lets the peripheral clocks work with the central clock, which is the proposed reason early time-restricted eating improved insulin sensitivity and pressure beyond the effect of eating less.
Who this may not transfer to:The controlled-feeding crossover trial enrolled men only; the larger weight-loss trial enrolled both men and women. Results in women for the pure timing effect are less established.
The studies · 2
Sutton EF, Beyl R, Early KS, Cefalu WT, Ravussin E, Peterson CM. Early Time-Restricted Feeding Improves Insulin Sensitivity, Blood Pressure, and Oxidative Stress Even without Weight Loss in Men with Prediabetes. · Cell Metab 2018;27(6):1212-1221.e3
Jamshed H, Steger FL, Bryan DR, et al. Effectiveness of Early Time-Restricted Eating for Weight Loss, Fat Loss, and Cardiometabolic Health in Adults With Obesity: A Randomized Clinical Trial. · JAMA Intern Med 2022;182(9):953-962
When It Goes Wrong
Trouble appears when the timing signals conflict, and the effects have been measured from the laboratory out to million-person cohorts.
Forced misalignment. Ten adults spent ten days on 28-hour days, so eating and sleeping fell across every circadian phase in turn. When behavior ran about 12 hours out of phase with the internal clock, glucose rose 6%, insulin 22%, mean arterial pressure 3%, leptin fell 17%, and the cortisol rhythm reversed completely; three of eight showed post-meal glucose in the prediabetic range while misaligned, in people whose glucose was normal when aligned. Ten people over ten days cannot show what a career of night shifts does. It does show that circadian misalignment itself, with food and sleep duration held constant, moves metabolic measurements in the direction the epidemiology points.
Social jetlag. The measurable version is the gap between mid-sleep on work days and mid-sleep on free days, which is what you correct by sleeping in on Saturday. In a database of about 65,000 people, social jetlag tracked with higher body mass index over and above sleep duration, and among people already at a BMI of 25 or more, each hour of it was associated with roughly 33% higher odds of sitting in a higher weight class. That association is cross-sectional and measured in people who mostly chose neither their work schedule nor their chronotype.
Evening chronotype and mortality. In UK Biobank, 433,268 adults followed a mean 6.5 years, definite evening types had a hazard ratio for death from any cause of 1.10 (95% CI 1.02 to 1.18) after wide adjustment. In the Finnish Twin Cohort, 23,854 people followed 37 years, the adjusted hazard ratio was 1.09 (1.01 to 1.18), and it fell away mainly through smoking and alcohol: among non-smokers who drank lightly or not at all, evening types showed no excess mortality. Those authors concluded there is little or no independent contribution of chronotype to mortality. Evening preference travels with smoking, drinking, shorter sleep and shift work, and the current data cannot separate the clock from those habits.
Shift work supplies the hardest outcome data on circadian disruption in humans, and it is also the most confounded. The IARC working group that met in June 2019 classified night shift work in Group 2A, probably carcinogenic to humans. The descriptors behind it usually get dropped: limited evidence in humans, with positive associations for cancers of the breast, prostate, colon and rectum; sufficient evidence in animals; and strong mechanistic evidence in experimental systems. In IARC's vocabulary "limited" means chance, bias and confounding could not be ruled out, and a Group 2A rating is a hazard identification that carries no estimate of how much risk a given job involves.
The largest test of the human half found the opposite. Pooling three prospective cohorts with seven previously published ones, about 1.4 million women and 4,660 breast cancers among shift workers, the relative risk of breast cancer with any night shift work was 0.99 (95% CI 0.95 to 1.03), including for long-term shift work, and the authors concluded night shift work has little or no effect on breast cancer incidence. That result sits inside the same literature IARC assessed, and it is why the human evidence is graded limited. The classification rests mainly on animal and mechanistic work, while the largest human breast-cancer analysis was flat.
The cardiometabolic associations are more consistent. Across 34 studies and 2,011,935 people, shift work tracked with myocardial infarction (RR 1.23), coronary events (RR 1.24) and ischemic stroke (RR 1.05), with no association with mortality. Across 226,652 people, shift work tracked with diabetes at an odds ratio of 1.09, higher in men (1.37) than in women (1.09) and higher for rotating shifts. None of these designs can separate the clock from the life around it: shift work travels with lower socioeconomic position, more smoking, shorter and more fragmented sleep, irregular eating and less daylight.
What This Means For You
Three inputs have measured circadian effects: the timing of light, the timing of sleep, and the timing of food. The practical shape of each is on its own page, and it follows from the mechanism described above.
- Light is the strongest lever because the phase response curve is steepest around the hours near waking, so daylight in the first part of your day moves the clock most for the least exposure. The free version outdoors is many times brighter than any indoor light, and a bright lamp is the fallback for a dark morning or a night shift.
- A fixed wake time delivers that light signal at the same clock phase every day, which is how the clock locks on; the rise time does more of that work than the bedtime does. It costs nothing and is the single change most sleep advice is built around.
- Front-loading food acts on the organ clocks in the liver, gut and fat, the ones a meal can move while light holds the master clock steady, so shifting the bulk of your calories toward the morning is the lever once light and sleep timing are set. It changes when you eat, not what you buy.
Go Deeper
The pages that act on this timing system:
- Morning light, which uses the advancing half of the phase response curve.
- Evening light and screens, the same curve on its delaying side, where the melanopic thresholds live.
- Sleep regularity, the steady sleep and wake time the clock locks onto.
- The sleep environment, where darkness, temperature and sound act on the clock.
- Insomnia, where a timing problem has to be told apart from the other things that keep people awake.
- Sleep restriction and stimulus control, which works on sleep pressure and conditioning alongside timing.
- Time-restricted eating, which acts through the food entrainment route described above.
- Insulin and glucose, for what the time of day does to a meal.
The Chinese Medicine View
Common Questions
Is the human body clock really 25 hours long?
No. Measured under controlled lighting, the intrinsic period averages 24.18 hours, and across 157 adults the spread was about 12 minutes either side. The 25-hour figure came from early isolation studies in which participants switched their own room lights on and off, so the light they chose was shifting the clock being measured. The correction was published in 1999.
Why does morning light wake me up while evening light keeps me up?
Because the clock's response to light flips sign at your core body temperature minimum, which falls about two hours before you normally wake. Light after that point advances the clock and pulls sleepiness earlier; light before it delays the clock and pushes sleep later. The only difference between the two is the hour the light arrives.
Does eating late really shift my body clock?
It shifts some of them. When ten men delayed all their meals by five hours, their plasma glucose rhythm delayed by 5.69 hours and a clock gene in their fat tissue delayed by about an hour, while melatonin and cortisol did not move. Meals set the clocks in liver, gut and fat; light sets the master clock in the brain.
Are teenagers who sleep late just badly disciplined?
No. Chronotype gets later through adolescence and reaches its latest point at 18.4 years in girls and 19.2 in boys, then moves earlier for the rest of life. That pattern shows up in 53,689 time-use diaries and it is the reason school start times have been argued about for two decades. It reverses on its own.
Does night shift work cause cancer?
IARC classifies night shift work as probably carcinogenic, Group 2A, and grades the human evidence limited, meaning chance, bias and confounding could not be ruled out. The largest analysis, about 1.4 million women, found a relative risk for breast cancer of 0.99. The animal and mechanistic evidence is what carries the classification. The cardiometabolic associations with shift work are more consistent than the cancer ones.
Explore Related
Other pages this one connects to, by the evidence they share, the outcomes they touch, and the ground they cover.
Pages that lead here: Insomnia · Morning Light
All 34 sources on this page independently checked and cross-referenced.
Thomas Dehli, Founder & Editor, Sacred Lotus
Sacred Lotus has published Chinese medicine reference material since 2001. Integrative pages are held to the same standard as the herb and formula library: cite the source, grade the claim at its real strength, and say where the research has not looked. This page is educational and it is not medical advice. Last reviewed and updated August 8, 2026.
Evidence strength
How confidently the research supports a claim. Strength describes the evidence, not our endorsement.