Skip to content
The Longevity Archive
browse 192
Extended12 human studies cited

72-Hour Fast

3 days without food · Once every few months in most accounts, and once as a diagnostic procedure

Written by Reviewed Sep 2026

Also called: Three day fast, The diagnostic fast

Three days, and the best-resolved duration in human fasting physiology, because 72 hours is a routine supervised diagnostic procedure in endocrinology. It is also the only duration with a muscle autophagy measurement, and that measurement is more ambiguous than it is usually reported to be.

Overview

The 72 hour fast has more human measurement behind it than any other extended duration, for an unglamorous reason: the supervised 72 hour fast is a decades-old diagnostic procedure used to investigate hypoglycaemia, so a great deal of biochemistry has been collected from healthy people completing one under observation.

What that produced is the best-resolved series in the field. Six healthy men sampled at 12, 18, 24, 30, 42, 54 and 72 hours with stable isotope tracers. Thirty-four adults with beta-hydroxybutyrate measured every 6 hours. Brain ketone concentrations measured directly by magnetic resonance spectroscopy. Muscle biopsies with forearm tracer kinetics at exactly 72 hours.

It is also where the most repeated fasting claim on the internet lives, that 72 hours regenerates the immune system. That traces to a 2014 paper whose stem cell regeneration, reversal of chemotherapy-induced immunosuppression and reversal of age-related myeloid bias were all measured in mice, and whose human component the paper's own abstract describes as preliminary data on the protection of lymphocytes from chemotoxicity in fasting patients.

How it works

Three days without calories, with water. In the diagnostic version this is done as an inpatient with 6-hourly blood sampling and a defined stopping rule, which is the setting that produced most of the numbers on this page. In the oncology dose-escalation the 72 hour version was 48 hours before chemotherapy plus 24 hours after.

What the supervised versions have in common is that someone is measuring. In a chart review of 768 medically supervised water-only fasting visits with a median length of 7 days, the two serious adverse events were a dehydration event on fasting day 3 and a grade 4 hyponatraemia on fasting day 9, both caught because a clinician was present.

Community practice at this duration is a full electrolyte routine, sodium from the start of the fast rather than in response to symptoms, with potassium and magnesium added. The specific amounts people use are far above the more cautious guides, and there is no randomised evidence establishing an electrolyte protocol for voluntary fasting at any duration. The one place a community practice can plausibly cause harm is concentrated potassium taken quickly, which can cause arrhythmia and gut injury and is far riskier for anyone with reduced kidney function or on an ACE inhibitor, ARB or potassium-sparing diuretic. The community itself says so.

What 3 days actually reaches

The tracer series in 6 healthy men gives the shape of the whole fast. Glucose fell from 5.58 ± 0.08 to 4.14 ± 0.10 mmol/L (100.5 to 74.6 mg/dL), about 25%. Glucose rate of appearance fell from 11.0 ± 0.4 to 8.3 ± 0.3 micromol/kg/min. Insulin fell from 64.6 ± 12.9 to 30.1 ± 7.9 pmol/L (p < 0.001), roughly half, with 70% of that decline complete inside the first 24 hours. Glycerol and palmitic acid release both roughly doubled, 2.08 ± 0.22 to 4.36 ± 0.36 and 1.63 ± 0.20 to 3.26 ± 0.40 micromol/kg/min, with 60% of the rise between 12 and 24 hours. The authors' own conclusion is that early lipid mobilisation tracks the fall in insulin and is not driven by glucose concentration or glucose kinetics. Ketones: median changes of 0% from 12 to 18 hours, +333% from 18 to 36, +210% from 36 to 54 and +167% from 54 to 72, measured every 6 hours in 34 adults. Using the existing diagnostic criterion, 74% of people whose fast was negative crossed a beta-hydroxybutyrate of 2.7 mmol/L before the 72 hour mark. The brain measurement is the most direct evidence of the fuel switch anyone has in a living person. Occipital beta-hydroxybutyrate by 4-tesla magnetic resonance spectroscopy went from 0.05 ± 0.05 mmol/L non-fasted to 0.60 ± 0.26 after the second day and 0.98 ± 0.16 after the third, with brain lactate rising from 0.69 ± 0.17 to 1.47 ± 0.22 mmol/L, and plasma and brain correlating at r = 0.86. Autophagy is where this duration earns its place and also where it needs the most care. In 8 healthy men at 72 hours, skeletal muscle mTOR Ser2448 phosphorylation fell about 50%, with reduced phosphorylation of ULK1 Ser757, 4EBP1 Thr46 and rpS6 Ser235/236, and LC3B-II rose about 30%. p62 also rose about 10% in the same biopsies, and p62 is degraded by autophagy, so it should fall if flux increased. The authors state directly that this makes interpretation of autophagic flux problematic and that they cannot exclude autophagosome accumulation and autophagy inhibition rather than induction. In a separate 72 hour water-only study, sorted CD45-positive leukocytes showed transcriptomic and proteomic upregulation of autophagy machinery, with no flux assay and no control group. That is the human evidence: an upstream signal that moves consistently, a downstream marker that moves ambiguously, and no flux measurement at this duration. It is enough for this page to list autophagy as a mechanism the window engages and not enough to say by how much. No protocol here claims mitophagy, because no human trial has measured mitophagy in response to any fast.

Fig. 1 · what has been measured, hour by hour
  1. 0h to 4h

    Still absorbing

    Burning
    Glucose from the meal just eaten, with insulin high enough to hold fat release down
    • Insulin is at its post-meal peak, and while it is there, adipose tissue lipase stays suppressed and fat is being stored rather than released.
    • The hour count on most fasting charts starts at the last bite. Absorption of a mixed meal is still running here, so the clock and the physiology are not yet the same thing.

    Read the biologyInsulin and insulin sensitivity during a fast

    Measured in people
    The human hour-by-hour fasting series begin later than this. Klein 1993 took its first sample at 12 hours in 6 healthy men, and Rothman 1991 reports its first gluconeogenesis figure as an average across the first 22 hours. This row is the fed state those measurements are compared against, and no extended-fasting measurement exists inside it.

  2. 4h to 12h

    Post-absorptive

    Burning
    A mix of remaining glucose and fatty acids, with the liver covering the brain's glucose demand
    Blood glucose
    5.58 ± 0.08 mmol/L (100.5 mg/dL) at the 12 hour mark, in 6 healthy men
    Ketones (BHB)
    Total ketone bodies around 0.20 mmol/L in overnight-fasted subjects. That figure is acetoacetate plus beta-hydroxybutyrate, not BHB on its own.
    Insulin
    64.6 ± 12.9 pmol/L (9.3 microunits/mL) at 12 hours, in the same 6 men
    • Glycerol release runs at 2.08 ± 0.22 and palmitic acid release at 1.63 ± 0.20 micromol/kg/min at 12 hours. Those are the numbers everything later in the fast is measured against.
    • Gluconeogenesis is already the majority contributor to glucose production rather than a reserve waiting to be called on, which is the first place the popular switch model breaks.

    Read the biologyInsulin and insulin sensitivity during a fastGlycogen depletion and gluconeogenesis

    Measured in people
    Klein 1993 measured glucose, insulin and stable-isotope lipid kinetics at 12 hours in 6 healthy men. Féry 1983 measured total ketone body turnover in overnight-fasted subjects and put the baseline at 0.20 mmol/L. Rothman 1991 measured hepatic glycogen serially by 13C nuclear magnetic resonance through a 68 hour fast in healthy adults and calculated that gluconeogenesis supplied 64 ± 5% of total glucose production across the first 22 hours.

  3. 12h to 18h

    Ketones have not moved yet

    Burning
    Fatty acids rising, with glucose production increasingly made rather than released from store
    Ketones (BHB)
    Median change of 0% between 12 and 18 hours, sampled every 6 hours in 34 adults
    • Beta-hydroxybutyrate does not move in this window. In 34 adults sampled every 6 hours through the standard diagnostic 72 hour fast, the median change from 12 to 18 hours was zero.
    • Gluconeogenesis is supplying 64 ± 5% of glucose production across the first 22 hours, confirmed by an orthogonal method at about 47% by 14 hours. Glucose production does not switch from glycogen to fat at a threshold hour.
    • Whole blood LC3A messenger RNA was 22 ± 5% higher at 18 hours than in a 12 hour control arm in 11 overweight adults (p = 0.001). In the same subjects MTOR expression also rose, by 9 ± 3%.

    Read the biologyGlycogen depletion and gluconeogenesisKetogenesis and beta-hydroxybutyrateAutophagy

    Measured in people
    Service 2005 measured beta-hydroxybutyrate every 6 hours in 34 adults whose 72 hour fasts were negative for insulinoma. Rothman 1991 measured the gluconeogenic share by 13C NMR. Jamshed 2019 measured whole blood messenger RNA in an 11-person 4 day randomised crossover of an 08:00 to 14:00 eating window against 08:00 to 20:00. That is one gene's expression in blood cells, not autophagic flux and not muscle. No study in any species has tested a 16 hour timepoint for autophagy.

  4. 18h to 1d

    The steepest hours for fat release

    Burning
    Fatty acids and glycerol, with ketone production starting its climb
    Ketones (BHB)
    Climbing. The median rise from 18 to 36 hours was +333%, the steepest proportional stretch of a 72 hour fast.
    Insulin
    70% of the entire decline seen across a 72 hour fast has already happened by 24 hours
    • The largest single interval increase in lipolysis across a whole 72 hour fast falls between 18 and 24 hours, and 60% of the total rise in lipid kinetics happens between 12 and 24 hours. The authors attribute that to the falling insulin rather than to any change in glucose.
    • Growth hormone was significantly higher after a 24 hour water-only fast than after a fed day in a randomised crossover of 30 healthy adults (p = 1.1 x 10^-4). The size of that rise is not in the accessible record and is not published here.
    • Haemoglobin, red cell count and haematocrit all rose together at 24 hours in the same trial. That combination is the signature of plasma volume contraction, not of new red cells.
    • 24 hour energy expenditure measured in a whole-room calorimeter fell during 24 hour fasting in 20 volunteers, with the size of the fall varying between individuals.

    Read the biologyInsulin and insulin sensitivity during a fastThe growth hormone responseKetogenesis and beta-hydroxybutyrateAutophagy

    Measured in people
    Horne 2013 randomised 30 apparently healthy adults to a 24 hour water-only fast or a day of usual eating in crossover and measured growth hormone, blood counts, cholesterol, triglycerides, bicarbonate and weight. Klein 1993 measured the lipid kinetics by stable isotope tracer. Hollstein 2020 measured 24 hour energy expenditure by whole-room calorimetry in 20 volunteers; the exact percentage decrease is not in the accessible record and is not published here. On autophagy, a 24 hour fast raised p62/SQSTM1 in the vastus lateralis of 50 women, which is movement in the direction of less autophagic degradation rather than more.

  5. 1d to 36h

    Thyroid drops before metabolic rate does

    Burning
    Fat, with the remaining glucose built from glycerol, lactate and amino acids
    Ketones (BHB)
    1.308 ± 1.053 mmol/L after day 1 in 13 men beginning a 10 day water fast, from a baseline of 0.177 ± 0.044. The spread on that mean is as wide as the mean.
    • Gluconeogenesis supplies 82 ± 5% of glucose production between 22 and 36 hours.
    • Serum T3 fell from 2.30 ± 0.06 to 1.84 ± 0.03 nmol/L by 30 hours in 8 healthy men (p < 0.01), TSH fell below 1 mU/L and the nocturnal TSH peak was abolished. Serum cortisol did not differ across conditions.
    • Thyroid suppression therefore begins inside the first 30 hours, well before any measurable fall in resting metabolic rate.
    • Sodium excretion has not turned yet. In 9 obese women fasting with pre-fast salt intake maintained, urinary sodium began to rise after roughly 48 hours and then exceeded intake.

    Read the biologyGlycogen depletion and gluconeogenesisKetogenesis and beta-hydroxybutyrateAutophagy

    Measured in people
    Rothman 1991 measured the gluconeogenic share by 13C NMR in healthy adults. Hugues 1984 fasted 8 healthy men for 30 hours with serial thyroid sampling. Dai 2022 measured blood beta-hydroxybutyrate daily in 13 men fasting in a controlled facility. Sigler 1975 collected urine in 3 hour blocks from 9 obese women on a metabolic ward. On autophagy, the only human study with several within-fast timepoints took vastus lateralis biopsies at 2, 12, 24 and 36 hours: LC3I, LC3II and p62 all fell, in untrained subjects only, which the authors summarise as skeletal muscle autophagy being only modestly affected by 36 hours of fasting.

  6. 36h to 2d

    Growth hormone multiplies, IGF-1 does not follow

    Burning
    Fat, with ketones now carrying a real share of the load and glucose production almost entirely gluconeogenic
    Ketones (BHB)
    Median +210% between 36 and 54 hours in 34 adults
    • 24 hour growth hormone production rose from 78 ± 12 to 371 ± 57 micrograms per litre of distribution volume on the second day of a fast in 9 normal men (p = 0.0001), a five-fold rise. Secretory bursts went from 14 ± 2.3 to 32 ± 2.4 per 24 hours and the mass per burst from 6.3 ± 1.2 to 11 ± 1.6 micrograms.
    • Growth hormone half-life was unchanged at 18 ± 2.2 against 20 ± 1.5 minutes (p = 0.47), so this is genuine secretion rather than slower clearance.
    • Serum IGF-1 was unchanged at 56 hours. It does not fall until around day 5, when it dropped from 1.31 ± 0.22 to 0.77 ± 0.18 U/mL in 6 men.
    • Gluconeogenesis supplies 96 ± 1% of glucose production from 36 hours onward.
    • Under an insulin clamp at 48 hours, whole-body glucose disposal fell from 39.8 ± 4.6 to 24.1 ± 2.1 micromol/kg/min (p < 0.01) and glucose oxidation collapsed from 21.8 ± 1.3 to 3.9 ± 1.4 (p < 0.001), while non-oxidative disposal was unchanged. That is a change in fuel selection, not glucose intolerance in the diabetic sense.
    • 24 hour energy expenditure in a respiration chamber was 10.9% lower on day 2 of a fast than in a fed condition at energy balance (9.8 ± 0.2 against 11.0 ± 0.4 MJ, p < 0.01).

    Read the biologyThe growth hormone responseGlycogen depletion and gluconeogenesisInsulin and insulin sensitivity during a fast

    Measured in people
    Hartman 1992 sampled serum growth hormone every 5 minutes for 24 hours in 9 normal men, on a fed day and on the second day of a fast, and applied multi-parameter deconvolution; IGF-1 was measured at 56 hours. Ho 1988 measured somatomedin C through a 5 day fast in 6 men. Mansell 1990 clamped 6 non-diabetic men at 48 hours of starvation. Andriessen 2023 put 12 healthy lean males in a respiration chamber for 60 continuous hours in a randomised crossover against a fed condition. No human study has measured autophagy in any tissue between 36 and 72 hours, which is why autophagy is not among the mechanisms linked from this row.

  7. 2d to 3d

    Ketones reach the brain

    Burning
    Fat and ketone bodies, with the brain now taking a measurable share of its fuel as beta-hydroxybutyrate
    Blood glucose
    4.14 ± 0.10 mmol/L (74.6 mg/dL) at 72 hours in 6 healthy men, about 25% below the 12 hour value
    Ketones (BHB)
    74% of people whose diagnostic fast was negative exceeded 2.7 mmol/L before the 72 hour mark. The median rise from 54 to 72 hours was +167%.
    Insulin
    30.1 ± 7.9 pmol/L (4.3 microunits/mL) at 72 hours, roughly half the 12 hour value (p < 0.001)
    • Brain beta-hydroxybutyrate, measured directly in the occipital lobe by 4-tesla magnetic resonance spectroscopy, rose from 0.05 ± 0.05 mmol/L non-fasted to 0.60 ± 0.26 after the second day and 0.98 ± 0.16 after the third. Brain lactate rose from 0.69 ± 0.17 to 1.47 ± 0.22 mmol/L.
    • Plasma and brain beta-hydroxybutyrate correlated at r = 0.86, with a brain-to-plasma slope of 0.26.
    • Across a 48 hour fast, triglyceride accumulated in the livers of men and in the muscles of women, while whole-body and hepatic glucose and oxidative responses were identical between the sexes.
    • The two energy expenditure measurements diverge here and both are real. Hood-measured resting expenditure rose 14.1% by day 3 in 11 lean subjects, while chamber-measured 24 hour expenditure was still 5.5% below the fed condition on day 3 (10.3 ± 0.3 against 10.9 ± 0.3 MJ, p < 0.01). The resting component can rise while the whole day falls.
    • The 72 hour supervised fast is a decades-old routine diagnostic procedure in endocrinology, which is the strongest available evidence that healthy people tolerate this duration under observation.

    Read the biologyKetogenesis and beta-hydroxybutyrateInsulin and insulin sensitivity during a fast

    Measured in people
    Klein 1993 sampled 6 healthy men at 12, 18, 24, 30, 42, 54 and 72 hours with stable isotope tracers and indirect calorimetry. Service 2005 measured beta-hydroxybutyrate every 6 hours in 34 adults. Pan 2000 measured brain beta-hydroxybutyrate and lactate by magnetic resonance spectroscopy in healthy adults. Browning 2012 measured liver and muscle triglyceride by spectroscopy across 48 hours in healthy men and women. Zauner 2000 measured resting expenditure by hood in 11 lean subjects and Andriessen 2023 measured 24 hour expenditure in a chamber in 12 lean men, and the two are measuring different quantities rather than contradicting each other.

  8. 3d onward

    Adapted, and the risk moves to the refeed

    past this window
    Burning
    Fat and ketones, with glucose production splitting between liver and kidney as the fast lengthens
    Blood glucose
    3.5 ± 0.5 mmol/L (63.1 mg/dL) on day 4 in 11 lean adults, and 3.2 ± 0.2 mmol/L (57.7 mg/dL) on day 5 in 6 men
    Ketones (BHB)
    Production begins to plateau after about 5 days. Across a 21 day fast in 13 volunteers, blood ketones went from 0.1 ± 0.04 to 6.61 ± 1.25 mmol/L.
    Insulin
    Reaches its plateau by day 3 in obese subjects fasting for weeks, while plasma glucagon rises about two-fold to its own peak on the same day
    • In 8 healthy men at 72 hours, skeletal muscle mTOR Ser2448 phosphorylation fell about 50% and LC3B-II rose about 30%. p62 rose about 10% in the same biopsies, and the authors state that this makes autophagic flux impossible to read from these markers alone.
    • Continuous glucose monitors on 12 healthy adults through a 7 day water-only fast recorded time below 70 mg/dL rising from 3.0 ± 7.1% to 66.0 ± 25.7% by day 5, with minimum daily glucose falling from 76 ± 14 to 50 ± 7 mg/dL, and no symptoms of hypoglycaemia reported. Symptom-based self-monitoring stops working at this point.
    • Serum uric acid roughly doubled by day 7, from 5.9 ± 0.4 to 12.5 ± 1.0 mg/dL in 15 obese patients, then fell back to 7.7 ± 1.3 by day 28 while ketosis continued. Urate clearance fell proportionally more than creatinine clearance, so this is a specific tubular handling change.
    • Resting energy expenditure fell 20.3 ± 11.13% across a 21 day complete fast in 13 volunteers, which is where the eventual metabolic decline shows up.
    • Urinary nitrogen excretion decreases progressively for about 4 weeks and then holds at roughly 3 to 6 g per day. Nitrogen loss per kilogram of weight lost is inversely related to body fat, roughly 20 g of nitrogen per kg in non-obese people against roughly 10 g per kg in those carrying 50 kg or more of fat.
    • Mineral balance runs hundreds of milligrams per day negative while serum phosphorus and magnesium stay unchanged. That dissociation is why the deaths in the case literature cluster in the refeeding period rather than in the fast.

    Read the biologyKetogenesis and beta-hydroxybutyrateAutophagyRefeeding, the leucine threshold and refeeding syndrome

    Measured in people
    Vendelbo 2014 took muscle biopsies from 8 healthy men after 72 hours with forearm tracer kinetics. Kolnes 2026 monitored 12 healthy adults continuously through a 7 day water-only fast. Fox 1976 measured serum urate and urate clearance through 28 days in 15 obese patients. Dai 2024 measured resting energy expenditure and blood ketones across a 21 day fast in 13 volunteers. Owen 1969 measured urinary nitrogen fractions in 11 obese subjects through 5 to 6 weeks, and Forbes 1979 established the scaling of nitrogen loss with body fat. Licata 1981 measured mineral balance directly over 40 days on a metabolic ward and found mean daily balances of -104 mg calcium, -48 mg magnesium and -363 mg phosphorus while serum phosphorus and magnesium did not change at all.

Human evidence

The best-resolved tracer series in the field, a 6-hourly ketone series in 34 adults, direct brain spectroscopy, muscle biopsies with forearm kinetics at exactly 72 hours, a leukocyte multi-omics study, and the oncology dose-escalation that used this duration as its longest arm.

  • Klein 1993, 6 healthy men sampled at 12, 18, 24, 30, 42, 54 and 72 hours with stable isotope tracers and indirect calorimetry. Glucose 5.58 ± 0.08 to 4.14 ± 0.10 mmol/L; glucose rate of appearance 11.0 ± 0.4 to 8.3 ± 0.3 micromol/kg/min; insulin 64.6 ± 12.9 to 30.1 ± 7.9 pmol/L (p < 0.001) with 70% of the decline inside the first day; glycerol and palmitic acid release both roughly doubling (American Journal of Physiology, 1993).
  • Service 2005, beta-hydroxybutyrate every 6 hours through the standard 72 hour diagnostic fast in 34 adults whose fasts were negative for insulinoma, alongside 21 with surgically confirmed insulinoma. Median changes 0%, +333%, +210% and +167% across the four intervals, and 74% of those with a negative fast exceeded 2.7 mmol/L before 72 hours (Journal of Clinical Endocrinology and Metabolism, 2005).
  • Pan 2000, occipital beta-hydroxybutyrate by 4-tesla magnetic resonance spectroscopy in healthy adults: 0.05 ± 0.05 mmol/L non-fasted, 0.60 ± 0.26 after two days, 0.98 ± 0.16 after three. Brain lactate 0.69 ± 0.17 to 1.47 ± 0.22 mmol/L. Plasma to brain correlation r = 0.86, slope 0.26 (Journal of Cerebral Blood Flow and Metabolism, 2000).
  • Vendelbo 2014, 8 healthy men studied postabsorptive and after 72 hours with forearm amino acid tracer kinetics and muscle biopsies. mTOR Ser2448 phosphorylation fell about 50%, with ULK1 Ser757, 4EBP1 Thr46 and rpS6 Ser235/236 all significantly down; LC3B-II rose about 30% and p62 rose about 10%. The authors state that the p62 rise makes autophagic flux interpretation problematic (PLoS One, 2014).
  • Bak 2016, 9 lean and 9 obese subjects compared at 12 and 72 hours with muscle and subcutaneous adipose sampling. mTOR signalling fell and ULK1 protein and messenger RNA rose. The adipose measures were lipolytic rather than autophagy markers (American Journal of Physiology, Endocrinology and Metabolism, 2016).
  • Qian 2021, 72 hour water-only fasting with sorted CD45-positive leukocytes, 57 subjects for biochemistry with multi-omics quality control passing in 4. Transcriptomic and proteomic upregulation of autophagy machinery. No flux assay and no control group (Aging Cell, 2021).
  • Dorff 2016, 20 patients in cohorts fasting 24, 48 or 72 hours around platinum-based chemotherapy, the 72 hour arm being 48 hours before plus 24 after. Fasting-related toxicities were all grade 2 or below. COMET-assay leukocyte DNA damage was reduced at 48 hours or more but at p = 0.08, and the neutropenia trend was p = 0.17. IGF-1 fell 30%, 33% and 8% in the three cohorts after the first fast. No control arm ate normally (BMC Cancer, 2016).

What this does not tell you: The tracer series is 6 healthy young men and the muscle biopsy work is 8. The 34-adult ketone series is patients undergoing a diagnostic procedure rather than healthy volunteers choosing to fast, though their fasts were negative. The leukocyte multi-omics study had no control group and no flux assay, and its multi-omics quality control passed in 4 subjects. The oncology work is a 20-patient dose escalation with no normally eating control. No randomised trial has tested a 72 hour fast in healthy people against a control with any clinical endpoint.

Reading the research record

The claim attached to this duration is that 72 hours resets or regenerates the immune system, and it is worth tracing carefully because the underlying work is real.

It comes from a 2014 Cell Stem Cell paper, covered by a university press release in June 2014. The hematopoietic stem cell regeneration, the reversal of chemotherapy-induced immunosuppression and the reversal of age-related myeloid bias were all measured in mice, through reduced IGF-1 and PKA signalling. The paper's own abstract describes its human component as preliminary data on the protection of lymphocytes from chemotoxicity in fasting patients. The 72 hours is the pilot's chosen protocol duration, not a demonstrated threshold at which a human immune system regenerates, the word entire does not come from the source, and the human participants were people undergoing chemotherapy rather than healthy people seeking a reset. No human study has demonstrated hematopoietic stem cell regeneration after a 72 hour fast.

The mouse-to-human dose mismatch is quantified in a study that ran human and mouse blood side by side. Fasting mice for 48 hours causes approximately 20% weight loss; starving human volunteers for up to 4 days causes less than 2%. In that same experiment a 48 hour mouse fast produced LC3B lipidation across all leukocyte subpopulations, while a 4 day human fast produced no detectable in vivo response and, ex vivo, a response only in neutrophils. There is no published formula converting mouse fasting hours to human fasting hours.

The second thing to correct here is the switch model. Gluconeogenesis was already supplying 64 ± 5% of glucose production across the first 22 hours, confirmed by deuterated water at about 47% by 14 hours, and reaches 96 ± 1% only after 36 hours. Glucose production does not switch from glycogen to fat at a threshold. What changes across three days is the proportion, continuously.

What happens to muscle

No trial has measured body composition across a 72 hour fast, and the reason is that at three days a scan is mostly reporting hydration.

What has been measured at exactly this duration is the protein flux. In 8 healthy men, forearm net phenylalanine release rose after 72 hours and skeletal muscle mTOR phosphorylation fell about 50%, with the net muscle amino acid loss driven more by suppressed synthesis than by accelerated breakdown. That is a real negative balance and it is not the same as a measured loss of tissue.

The closest quantified answers come from 10 day studies. In 13 men through a complete water fast with DXA at baseline, fasting day 6 and recovery day 5, lean mass was down 9.2% at day 6 and back to baseline after 4 days of calorie restriction plus 5 days of refeeding, while fat mass ended 17.2% down. In 16 men on a 10 day modified fast, lean soft tissue loss of 3.53 ± 0.13 kg decomposed into 44% extracellular water, 14% glycogen with its bound water and 42% metabolically active lean tissue, which is 1.5 ± 0.1 kg or 25% of total weight loss. Together those are the strongest evidence that most acute lean mass loss during a short fast is water and glycogen rather than myofibrillar protein.

The functional test settles it more directly than any scan. 13 healthy young adults, 6 women and 7 men, completed a seven day water only fast with DXA and strength testing. Lean mass fell 4.6 +/- 0.3 kg, an 8.0% drop, and 1.7 +/- 0.5 kg of that came off the arms and legs. Fat mass fell 1.4 +/- 0.1 kg. Muscle glycogen halved. And maximal isometric and isokinetic strength were unchanged. A scan said they had lost 8% of their lean mass and their muscles were exactly as strong. Peak oxygen uptake did fall 13%, so endurance is the capacity that suffers, not force. The authors state directly that whether the protein degradation during fasting includes contractile proteins is uncertain.

The protein-sparing adaptation has not arrived by 72 hours. Protein oxidation measured as urinary nitrogen dropped 41 ± 7% by day 5 and then held stable through day 10, and plasma 3-methylhistidine, a marker of myofibrillar breakdown specifically, rose until day 5 before falling. Three days sits in the early proteolysis part of that curve.

And the scaling caveat again, because it changes who this applies to: nitrogen loss per kilogram of weight lost is roughly 20 g of nitrogen per kg in non-obese people against roughly 10 g per kg in people carrying 50 kg or more of fat. Almost all the classic prolonged starvation physiology was measured in obese subjects.

Doing it

Symptom-based monitoring has stopped working by day three
Continuous glucose monitors on 12 healthy adults through a 7 day water-only fast recorded time below 70 mg/dL rising from 3.0 ± 7.1% at baseline to 66.0 ± 25.7% by day 5, with minimum daily glucose falling from 76 ± 14 to 50 ± 7 mg/dL, and no symptoms of hypoglycaemia reported. In a healthy person that is adaptation to ketone fuel. In anyone on a glucose-lowering drug it is a lost warning system.
Sodium is leaving faster than it arrives
Sodium excretion begins rising after roughly 48 hours and then exceeds intake, peaking at day 4 at 109 ± 8 mmol/day on a constant 51 mmol/day intake, with a mean cumulative negative balance of 203 mEq in week one. Aldosterone cannot compensate: infused on day 4 it produced only about 40% of its pre-fast antinatriuretic effect. The one grade 4 event in the largest supervised water-only cohort was hyponatraemia.
Urate rises and it is not a hydration problem
Serum urate roughly doubles by day 7 and the rise scales with the depth of ketosis. The mechanism is competition between ketoacids and urate for the same proximal tubular transport, with urate clearance falling proportionally more than creatinine clearance, so drinking more water does not fix it. In 1,422 supervised subjects there was one gout attack, in a man already on allopurinol with frequent prior attacks.
Potassium is where community practice carries real risk
Some circulating protocols put daily potassium at or above 2,000 mg delivered in a drinkable liquid. Concentrated potassium taken quickly can cause arrhythmia and gut injury, and the risk is much higher with reduced kidney function or on an ACE inhibitor, ARB or potassium-sparing diuretic. The community's own guides say to avoid tablets above 99 mg and never to take it as a fast slug.
Training
There is no trial of resistance training during a 72 hour fast. Community practice is to keep walking, drop intensity and not attempt personal records, with lifting on days two and three described as noticeably weaker. What is measured at longer durations under supervision is that light activity was maintained and weight-bearing strength rose across a 10 day modified fast.
Plan the refeed before you start
The most consistently offered piece of advice in fasting communities is behavioural rather than physiological: buy and prepare the first food before the fast ends, so the decision is not made while hungry. What people report going wrong after a long fast is a large or carbohydrate-heavy first meal and the reactive over-eating that follows it.

Breaking the fast

This is the part of a 72 hour fast with the most published evidence behind it, and it is about the mechanism rather than the menu. During a fast, phosphate, potassium and magnesium are depleted from the body while serum values look normal, because the cells shrink their contents in parallel with total body stores. The cleanest demonstration is a 40 day metabolic ward study in 6 subjects: mean daily balances of -104 mg calcium, -48 mg magnesium and -363 mg phosphorus, with cumulative urinary losses exceeding intake by 58%, 75% and 500% respectively, while serum phosphorus and magnesium did not change at all. Reintroducing carbohydrate raises insulin, which drives glucose, phosphate, potassium and magnesium into cells simultaneously, and consumes thiamine as the cofactor for pyruvate dehydrogenase at the same moment. That is why the case literature clusters where it does. A death from lactic acidosis during refeeding after a 3 week fast. A death on the 7th or 8th day of realimentation after a 210 day fast, with gross fragmentation of cardiac myofibrils at necropsy. Gross fluid retention on refeeding in two of three crises in an inpatient series. In the 382 day fast, the lowest potassium of the entire episode, 2.9 mEq/L, occurred on day 6 of refeeding, and magnesium fell further in the first days of refeeding too. At three days in a well-nourished person, none of the formal risk thresholds are met: NICE treats little or no intake for more than 5 days as one of two criteria and more than 10 days as a single sufficient one, and ASPEN's bands are 5 to 6 days for moderate risk and more than 7 for significant risk. The realistic problems at 72 hours are bloating, cramping, diarrhoea, nausea and a heavy crash after a carbohydrate-heavy first meal, plus reactive over-eating over the next day. Community practice, offered as practice rather than evidence, is a cup of broth, a 30 minute wait, a second cup, then something small and fatty an hour or two later, then a small portion of cooked protein with a well-cooked vegetable, with a normal meal not until the next day, and electrolytes continued through the refeed rather than stopped at the first meal.

Citations

  1. Human2025
    Effects of seven days' fasting on physical performance and metabolic adaptation during exercise in humans

    Nature Communications

    13 healthy young adults, 6 women and 7 men, completed a seven day water only fast (15 enrolled, 2 discontinued). Body weight fell 5.8 +/- 0.3 kg. DXA lean mass fell 4.6 +/- 0.3 kg, an 8.0% decrease, with 1.7 +/- 0.5 kg of it from the extremities, a 6% reduction across arms and legs. Fat mass fell 1.4 +/- 0.1 kg. Muscle glycogen halved. Maximal isometric and isokinetic strength were UNCHANGED, while peak oxygen uptake fell 13% and endurance capacity dropped. Electron transport chain protein expression was unchanged, PDK4 rose 13-fold, and AMPK activity was unaffected, which the authors say challenges its proposed role in muscle protein degradation. The authors state that whether fasting protein degradation includes contractile proteins is uncertain.

  2. Human1993
    Progressive alterations in lipid and glucose metabolism during short-term fasting in young adult men

    American Journal of Physiology

    6 healthy men sampled at 12, 18, 24, 30, 42, 54 and 72 hours with stable isotope tracers and indirect calorimetry. Glucose 5.58 ± 0.08 to 4.14 ± 0.10 mmol/L (100.5 to 74.6 mg/dL); glucose rate of appearance 11.0 ± 0.4 to 8.3 ± 0.3 micromol/kg/min; insulin 64.6 ± 12.9 to 30.1 ± 7.9 pmol/L (p < 0.001) with 70% of the decline inside the first 24 hours; glycerol 2.08 ± 0.22 to 4.36 ± 0.36 and palmitic acid 1.63 ± 0.20 to 3.26 ± 0.40 micromol/kg/min. Early lipid mobilisation tracked the fall in insulin, not glucose kinetics.

  3. Human2005
    Increasing serum betahydroxybutyrate concentrations during the 72-hour fast: evidence against hyperinsulinemic hypoglycemia

    Journal of Clinical Endocrinology and Metabolism

    Beta-hydroxybutyrate measured every 6 hours through the standard 72 hour diagnostic fast in 34 adults whose fasts were negative for insulinoma, alongside 21 with surgically confirmed insulinoma. Median changes 0% from 12 to 18 hours, +333% from 18 to 36, +210% from 36 to 54 and +167% from 54 to 72. 74% of those with a negative fast exceeded 2.7 mmol/L before the 72 hour mark.

  4. Human2000
    Human brain beta-hydroxybutyrate and lactate increase in fasting-induced ketosis

    Journal of Cerebral Blood Flow and Metabolism

    Occipital beta-hydroxybutyrate measured directly by 4-tesla magnetic resonance spectroscopy in healthy adults: 0.05 ± 0.05 mmol/L non-fasted, 0.60 ± 0.26 after the second day, 0.98 ± 0.16 after the third. Brain lactate 0.69 ± 0.17 to 1.47 ± 0.22 mmol/L. Plasma and brain correlated at r = 0.86 with a brain-to-plasma slope of 0.26.

  5. Human2014
    Fasting increases human skeletal muscle net phenylalanine release and this is associated with decreased mTOR signaling

    PLoS One

    8 healthy men studied postabsorptive and after a 72 hour fast, with forearm amino acid tracer kinetics and muscle biopsies. mTOR Ser2448 phosphorylation fell about 50%, with ULK1 Ser757, 4EBP1 Thr46 and rpS6 Ser235/236 all significantly reduced. LC3B-II rose about 30% and p62 rose about 10%. Because p62 is itself degraded by autophagy, the authors state this makes autophagic flux interpretation problematic and that they cannot exclude autophagosome accumulation and autophagy inhibition rather than induction.

  6. Human2016
    Differential regulation of lipid and protein metabolism in obese vs. lean subjects before and after a 72-h fast

    American Journal of Physiology, Endocrinology and Metabolism

    9 lean and 9 obese subjects compared at 12 and 72 hours with muscle and subcutaneous adipose sampling. mTOR signalling fell and ULK1 protein and messenger RNA rose. The adipose measures were lipolytic rather than autophagy markers.

  7. Human2021
    Innate immune remodeling by short-term intensive fasting

    Aging Cell

    72 hour water-only fasting with sorted CD45-positive leukocytes, 57 subjects for biochemistry with multi-omics quality control passing in 4. Transcriptomic and proteomic upregulation of autophagy machinery. There was no flux assay and no control group, so this measures the expression of the machinery rather than the rate at which it runs.

  8. Animal2014
    Prolonged fasting reduces IGF-1/PKA to promote hematopoietic-stem-cell-based regeneration and reverse immunosuppression

    Cell Stem Cell

    The origin of the claim that 72 hours regenerates the immune system. The hematopoietic stem cell regeneration, the reversal of chemotherapy-induced immunosuppression and the reversal of age-related myeloid bias were all measured in mice. The paper's own abstract describes the human component as preliminary data on the protection of lymphocytes from chemotoxicity in fasting patients. No human study has demonstrated hematopoietic stem cell regeneration after a 72 hour fast.

  9. Human2017
    Metabolic effects of fasting on human and mouse blood in vivo

    Autophagy

    Human and mouse blood run in the same study, with 4 to 5 subjects per condition. No enhanced autophagy-associated lipidation was detectable in vivo in human leukocytes after up to 4 days of fasting; flux appeared only after ex vivo culture with leupeptin, and among human cell types only in neutrophils. The paper also quantifies the dose mismatch: 48 hours of fasting causes approximately 20% weight loss in mice and less than 2% in humans over 4 days.

  10. Human2016
    Safety and feasibility of fasting in combination with platinum-based chemotherapy

    BMC Cancer

    20 patients in cohorts fasting 24, 48 or 72 hours around platinum-based chemotherapy, the longest arm being 48 hours before plus 24 after. Feasibility was met and fasting-related toxicities were all grade 2 or below. Leukocyte DNA damage by COMET assay was reduced at 48 hours or more at p = 0.08, not significant, and the trend to less grade 3/4 neutropenia was p = 0.17. IGF-1 fell 30%, 33% and 8% in the 24, 48 and 72 hour cohorts after the first fast. No control arm ate normally.

  11. Human2026
    Marked Increases in Continuous Glucose Monitor-Detected Hypoglycemia During a Seven-Day Water-Only Fast in Healthy Men and Women

    Journal of Diabetes Science and Technology

    12 healthy adults, 7 men, mean age 29.7 and mean BMI 25.0, wearing continuous glucose monitors through a 7 day water-only fast. Time below 70 mg/dL rose from 3.0% ± 7.1% at baseline to 66.0% ± 25.7% by day 5 (P < .001) and minimum daily glucose fell from 76 ± 14 to 50 ± 7 mg/dL (P < .001). No symptoms of hypoglycaemia were reported, which makes symptom-based self-monitoring unreliable during an extended fast.

  12. Human1981
    Adverse effects of liquid protein fast on the handling of magnesium, calcium and phosphorus

    American Journal of Medicine

    6 obese subjects studied for 40 days on a metabolic ward. Cumulative urinary losses exceeded intake by 58% for calcium, 75% for phosphorus and 500% for magnesium, with mean daily balances of -104 mg calcium, -48 mg magnesium and -363 mg phosphorus, while serum phosphorus and magnesium did not change and serum calcium fell only 0.5 mg/dL. A normal serum magnesium or phosphate during a fast does not mean the store is intact, which is why refeeding is the dangerous phase.

  13. Human2022
    Effects of 10-Day Complete Fasting on Physiological Homeostasis, Nutrition and Health Markers in Male Adults

    Nutrients

    13 male adults in a controlled facility: 3 day baseline, 10 day complete water-only fast, 4 days of calorie restriction, 5 days of full recovery, with DXA at baseline, fasting day 6 and recovery day 5. Body weight fell 7.28 kg (9.8%). At fasting day 6, total fat mass was down 10.7% and lean mass 9.2%; by recovery day 5, fat mass was down 17.2% and lean mass had returned to baseline. Glucose reached a nadir of 3.508 ± 0.397 mmol/L on day 4. Serum sodium fell below the 137 mmol/L lower limit on day 9, and the authors conclude that a sodium chloride supplement should be considered.

What people report

These are uncontrolled self-reports, not evidence. They are here because they tell you what to expect and what to watch for, which the trial literature does not. They cannot tell you whether anything works.

  • Three days treated as the first genuinely extended fast and as a distinct threshold, with day two described as the wall and the highest quit point
  • Physical hunger substantially receding by day three, described as the community's most confidently held claim about extended fasting
  • Appetite persisting after physical hunger goes, with people describing constant thoughts about food alongside no stomach sensation, a distinction they make carefully
  • Acetone breath and a metallic taste, described as strong by day two or three
  • Orthostatic dizziness worse than at 48 hours, with near-universal advice to stand up in stages
  • Persistent cold, with people adding layers indoors
  • A calm, quiet, emotionally flat focus on day three, reported alongside a minority who describe day three fog, with no explanation offered for the split
  • Heart palpitations treated as one of the few signals where the community's default advice is to take electrolytes and, if they persist, stop rather than push through

Sources: Uncontrolled self-report gathered from public fasting communities, with no control group, no denominator and no follow-up on those who stopped posting. The hunger item has a supervised counterpart pointing the same way: in a cohort of 1,422 people fasting 4 to 21 days on 200 to 250 kcal a day, 93.2% reported an absence of hunger. Presyncope and insomnia both appear in the supervised chart review, in 217 and 257 of 768 visits.

Who this is wrong for

  • Anyone taking an SGLT2 inhibitor, unless a prescriber has stopped it at least 3 days ahead, or 4 for ertugliflozin. The resulting ketoacidosis occurs at a normal-looking glucose, and a urine ketone strip is a poor test because it measures acetoacetate rather than beta-hydroxybutyrate. If nausea, vomiting, abdominal pain, air hunger or confusion appear, the relevant tests are a venous blood gas and a beta-hydroxybutyrate level, not a fingerstick.
  • Anyone on insulin, a sulfonylurea or a meglitinide without a prescriber adjusting the dose in advance, and not on the assumption that a dose cut is enough: 50% basal and 70% prandial reductions each still produced significant hypoglycaemia rates.
  • Anyone with type 1 diabetes outside a structured programme with continuous glucose monitoring and an insulin plan.
  • Anyone with primary adrenal insufficiency, where the cortisol response a fast normally calls for is a fixed oral dose that cannot rise.
  • Pregnancy and breastfeeding. There is no comparable data for multi-day fasting in either, and the glucose demand is continuous rather than intermittent.
  • Anyone with a history of an eating disorder or current loss-of-control eating. A large supervised fasting clinic excludes cachexia and anorexia nervosa outright.
  • Children and adolescents.
  • Anyone with a BMI under 18.5 or meeting the NICE refeeding risk criteria on the way in, because a low-BMI person enters a fast already depleted of intracellular phosphate, potassium, magnesium and thiamine.
  • Anyone with chronic kidney disease, and particularly on an ACE inhibitor, ARB or diuretic. In 131 fasting CKD patients, creatinine rose in 60.4% by day 7, associated with RAAS antagonist use (relative risk 2, P = 0.002), with major adverse cardiovascular events in six fasting patients against one non-fasting (P = 0.036).
  • Anyone with known cardiac disease outside supervision. Of the six fatal cases in the historical review of therapeutic fasting, two had severe pre-fast congestive heart failure that had initially appeared to improve with fasting and one had focal coronary artery stenosis at autopsy.
  • Anyone with established gout, and anyone taking lithium who will not be drinking freely.
  • Anyone with sickle cell trait, which is common, frequently unknown to the carrier, and for which dehydration is the specific documented stressor. A 29-year-old woman with an unremarkable history died suddenly during religious fasting, with red cell sickling in lungs, liver and spleen at postmortem and trait confirmed on electrophoresis. That is a single case report and is weighted as one.

Questions

Does a 72 hour fast reset the immune system?
The study that claim comes from measured stem cell regeneration, reversal of chemotherapy-induced immunosuppression and reversal of age-related myeloid bias in mice. Its own abstract describes the human component as preliminary data on the protection of lymphocytes from chemotoxicity in fasting patients, who were people undergoing chemotherapy. The 72 hours is the pilot's protocol length, not a demonstrated threshold, and no human study has shown hematopoietic stem cell regeneration after a fast.
Is autophagy actually elevated at 72 hours?
The upstream signal moves consistently: mTOR Ser2448 phosphorylation fell about 50% in the muscle of 8 healthy men, with ULK1 Ser757 dephosphorylated. The downstream marker is ambiguous: LC3B-II rose about 30%, but p62 rose about 10% in the same biopsies, and p62 should fall if flux increased. The authors say directly that they cannot exclude autophagosome accumulation and autophagy inhibition. A separate study found leukocyte autophagy machinery upregulated at 72 hours with no flux assay and no control group. So the window engages the pathway, and nobody has measured by how much.
How low does blood sugar go?
About 25% below the 12 hour value: 4.14 ± 0.10 mmol/L, or 74.6 mg/dL, at 72 hours in 6 healthy men. It keeps falling after that, to 3.5 ± 0.5 mmol/L on day 4 and 3.2 ± 0.2 on day 5. Continuous monitoring through a 7 day fast in 12 healthy adults recorded a minimum daily glucose of 50 ± 7 mg/dL by day 5 with no hypoglycaemia symptoms, because the brain has ketones. That silence is reassuring in a healthy person and dangerous in anyone on a glucose-lowering drug.
Is a 72 hour fast safe?
The supervised 72 hour fast is a routine diagnostic procedure in endocrinology, which is the strongest evidence available that healthy people tolerate this duration under observation. In a chart review of 768 supervised water-only fasting visits with a median length of 7 days, the highest-grade adverse event was grade 2 or lower in 555 visits and grade 3 in 212, with one grade 4 and no deaths. The two serious events were a dehydration event on fasting day 3 and a hyponatraemia on day 9, both in men aged over 70, and both were caught because a clinician was present.
How much muscle does three days cost?
Nobody has measured body composition across a 72 hour fast. Forearm net phenylalanine release rises at 72 hours, driven more by suppressed synthesis than by accelerated breakdown, so the balance is negative. The nearest quantified answers come from 10 day studies: 58% of the lean soft tissue loss in one was extracellular water and glycogen with its bound water, and in another lean mass measured by DXA had returned to baseline after refeeding while fat mass stayed down.