Fasting: my monthly 3-day water and broth fast

The short version

I do a 3-day water and broth fast once a month. No trial I found tested that exact schedule, so this page covers the nearest evidence, and it's a self-experiment, not a proven protocol.
By hour 72, six healthy men on water had glucose near 71 mg/dL and the ketone beta-hydroxybutyrate (BHB) near 2.3 mM, up from 0.1¹. In eight men, a 72-hour fast increased amino acid release from muscle².
A review of 5-to-20-day water fasts found blood pressure consistently fell, but about two-thirds of the weight lost was lean mass, and in the few trials that checked, the metabolic gains were gone 3 to 4 months after the fast³.
The best-known "periodic fasting" trial used a 5-day food-based diet, not a water fast, and measured markers, not disease⁴.
Valter Longo, who studies fasting, says that for people with diabetes, anorexia, or a disease they take drugs for, the doctor has to decide⁵, and a large fasting clinic's contraindications include pregnancy and anorexia nervosa⁶. I take retatrutide; how that fits is a question for my doctor, and the general risks are below.

The protocol: what the research used

This is what studies tested, not a prescription. My own version is in the last row.

Study	What was done
Lab studies of 72 hours	Water only; eight men (muscle)² and six men (glucose and ketones)¹
84-hour lab fast	Eleven healthy, lean volunteers⁷
Buchinger clinic cohort	4 to 21 days at roughly 200 to 250 kcal a day of juice and vegetable soup, plus 3 liters of water or herbal tea a day, in 1,422 people⁶
Fasting-mimicking diet	5 consecutive days a month for 3 months, 100 people⁴
Prolonged water-fast trials	5 to 20 days, reviewed³
Mine	3 days, once a month, water and broth (what I've told the site)

Why it works (or is claimed to)

Over 72 hours without food, insulin fell below the assay's detection limit in five of six men while free fatty acids and ketones rose¹, and BHB is a fuel the brain uses during starvation⁸. Stress hormones climb: in 11 lean volunteers fasting 84 hours, serum norepinephrine rose from 1,716 to 3,728 pmol/L by day 4 and resting energy expenditure went up (3.97 to 4.53 kJ/min by day 3) while glucose fell from 4.9 to 3.5 mmol/L⁷. The popular claims go further, to cell clean-up (autophagy) and regeneration^9,10, and that's where the evidence thins, as below.

What happens over 72 hours

Here are the measured values from the water-only 72-hour fast in six healthy men, compared with a baseline taken after a 12-hour fast¹:

Measure	Day 0	Day 3 (72 hours)
Glucose	85.1 mg/dL	71 mg/dL
Free fatty acids	0.6 mmol/L	1.1 mmol/L
BHB	0.1 mmol/L	2.3 mmol/L
Insulin	3.1 µU/mL	below the 2.0 µU/mL detection limit in 5 of 6 men

(Values are means; BHB varied between men, with a standard error of 0.5¹.) Nutritional ketosis is defined as blood BHB of at least 0.5 mM, and healthy people typically sit near 0.1 mM¹¹, so day 3 was well into ketosis. I follow keto, so my starting point is probably not theirs, and I didn't find a study that tracked a keto-adapted person through a 3-day fast. None of these studies used broth, and I didn't find one that tested what broth changes.

The same fast lowered glucose enough that continuous monitoring caught frequent hypoglycemic episodes in healthy men¹. In a separate 72-hour fast in eight men, net phenylalanine (an amino acid) release from the forearm rose, mTOR signaling (a growth switch) fell about 50%, and the insulin-stimulated rise in that signaling was blunted². An autophagy marker rose about 30%, but p62, a protein that autophagy breaks down, rose about 10% too, so the authors called the interpretation of autophagy problematic². A randomized crossover trial of a three-day water-only fast, with autophagy as its main outcome, has published its protocol; I found no results yet¹².

What the research says

Glucose, ketones, and stress hormones

Early These are small lab studies (6 and 11 people), but they agree that glucose falls within 72 to 84 hours and that stress hormones climb: glucagon and epinephrine in McDougal's six men, norepinephrine in Zauner's 11 volunteers^1,7. McDougal also measured free fatty acids and ketones rising. They measure physiology, not health outcomes.

Lean mass and muscle

Early The prolonged-water-fast review found about two-thirds of the weight lost in 5-to-20-day fasts was lean mass, which the authors flagged as a concern³. The 72-hour muscle study showed the pathway going the wrong way: more net amino acid release and less growth signaling². The numbers are from long fasts and from signaling markers; I did not find a measured muscle loss figure for a single 3-day fast, or for a monthly one. For anyone who trains, that is the gap that matters most.

Periodic fasting and health markers

Mixed The same review found blood pressure consistently fell, lipid results were inconsistent, and fasting glucose, insulin, insulin resistance, and HbA1c fell in people with normal glucose but did not change in people with type 1 or type 2 diabetes; in the few trials that checked, the benefits were gone 3 to 4 months after the fast, even when the weight loss held³. It is a narrative review, and the trials it covers ran 5 to 20 days, not 3.

The Buchinger cohort

Early In 1,422 people who fasted 4 to 21 days at a clinic in Germany, weight, abdominal circumference, and blood pressure fell, ketones rose, 93.2% reported no hunger, and adverse effects occurred in fewer than 1%⁶. It was uncontrolled, the clinic's physicians and nurses collected the data, and the study was funded through a company that develops the clinic's research, with several authors working for or paid by the clinic⁶. The diet was about 200 to 250 kcal a day of juice and vegetable soup under medical supervision, not water and broth alone, so I treat it as a safety signal for supervised fasting and no more⁶.

The fasting-mimicking diet

Moderate for markers, Early for anything bigger. In a randomized trial of 100 generally healthy people, three monthly 5-day cycles of the fasting-mimicking diet reduced body weight, trunk and total body fat, and blood pressure and lowered IGF-1, with no serious adverse effects; a post hoc analysis found larger changes in people at risk⁴. The diet is low in calories, sugars, and protein and high in unsaturated fats, so it supplies food and isn't my fast⁴. L-Nutra supplied the diet, the University of Southern California has licensed intellectual property behind it to L-Nutra, and two authors, Valter Longo among them, hold equity in the company, though they did not collect or analyze the data⁴. The longevity, regeneration, and cognition results come from yeast and mice, with only a pilot trial in people¹⁰.

What isn't shown

I didn't find a randomized trial of a 3-day water fast repeated monthly with disease or longevity outcomes. Autophagy in people after a 3-day fast is still being tested¹². Longo has said the cancer work started in mice, where the effects were very strong⁵, and mice aren't people. A 5-day diet that improves markers in 100 people doesn't tell me what a 3-day water fast does to my heart or muscle.

Electrolytes, uric acid, and refeeding

In the Buchinger cohort's full text, uric acid rose from 338.1 to 495.2 µmol/L on average, and one man already treated for gout had a gout attack⁶. Its adverse-event table lists three cases of low sodium (hyponatremia), three of cardiac arrhythmia, two of hypoglycemia, and one of low potassium; the text describes six mild cases of low sodium (lowest 127 mmol/L), all corrected with salt, and calls the arrhythmias low-grade and transitory⁶. Two people were hospitalized: a 75-year-old man with known coronary artery disease who had a heart attack on day 9, and a 67-year-old woman with vomiting, dizziness, and diarrhea on day 4⁶. Those fasts included 3 liters of water or herbal tea a day, a daily juice and soup, a medically supervised stay, and about four days of stepwise refeeding from 800 to 1,600 kcal a day⁶. A water-and-broth fast at home has none of that built in.

Refeeding syndrome is the risk when eating restarts. The American Society for Parenteral and Enteral Nutrition defines it as a 10% to 20% (mild), 20% to 30% (moderate), or over 30% (severe) drop in blood phosphorus, potassium, or magnesium within 5 days of reintroducing calories, with organ dysfunction or thiamin deficiency also marking severe cases¹³. That definition was written for hospitalized adults and children, I found no study of refeeding problems after a 3-day fast in a healthy adult, and how to break a fast is a question for my doctor.

Who shouldn't do it

People with diabetes, anorexia, or a disease they take drugs for. Longo says the doctor is the one who has to decide⁵.
Anyone on insulin or a sulfonylurea. A 72-hour water fast alone drove glucose low enough for frequent hypoglycemic episodes in healthy men¹, and the tirzepatide label warns that adding insulin or a secretagogue such as a sulfonylurea can raise the risk of hypoglycemia, including severe hypoglycemia¹⁴.
People on SGLT2 inhibitors. A case series described ketoacidosis or ketosis in nine people on these diabetes drugs, seven of them with type 1 diabetes, even without very high glucose, and said the drugs should be used in type 1 diabetes only with great caution, extensive counseling, and close monitoring¹⁵.
The Buchinger clinic's contraindications: cachexia, anorexia nervosa, advanced kidney, liver, or cerebrovascular insufficiency, dementia, pregnancy, and lactation⁶.
A history of gout (a signal, not a rule). Uric acid rose in the cohort, and a man already treated for gout had an attack⁶.
Heart disease (a signal, not a rule). Arrhythmia occurred in 3 of 1,422 supervised fasters, and the one heart attack was in a man with known coronary artery disease⁶.

Fasting while on glucose-lowering drugs

I take retatrutide; my retatrutide page covers the drug itself, and my blood sugar page covers diabetes drugs more broadly. This section is the general risk of fasting on drugs that lower glucose, with its sources, not a plan.

Low blood sugar. Fasting lowers glucose on its own¹. The label for tirzepatide, a related drug, says it can cause hypoglycemia; in its type 2 diabetes trial, hypoglycemia below 54 mg/dL ran 10.3% in people also taking a sulfonylurea versus 2.1% in those who weren't¹⁴.
Fluids and kidneys. The same label warns of gastrointestinal reactions, sometimes severe, and of acute kidney injury from volume depletion¹⁴. Put a fast on top of vomiting or diarrhea and dehydration is the obvious concern; that's my inference, not a tested result.
What I didn't find. I found no study of a multi-day fast in people taking retatrutide, tirzepatide, or semaglutide.

How this fits with retatrutide is a question for my doctor, and this page gives no plan.

Rhonda Patrick on fasting

She has covered fasting, but these are conversations and a topic page, not trials, and they don't mean she endorses this site. She is the interviewer on the Longo episode, not an author of his trials.

Valter Longo, 2016: the fasting-mimicking diet versus prolonged fasting, mouse results, and refeeding. In the transcript she says she wouldn't do a multi-day fast without some supervision, and Longo says people usually get a doctor's clearance first⁵.
Topic page, updated July 11, 2026: it lays out time-restricted eating, alternate-day fasting, prolonged fasting (over 48 hours, typically 48 to 96 hours in research), and the fasting-mimicking diet, and says periodic fasting may be more appropriate under a physician's guidance⁹.
Eric Verdin, 2017: ketogenic diets and aging, with intermittent fasting as an alternative¹⁶.

How I use it

I do a 3-day water and broth fast once a month. I also follow keto and take retatrutide. That's what I've told the site; the rest is blank on purpose.

Cautions

It's a self-experiment with no trial behind the exact schedule, and in longer fasts the marker gains faded within months³.
Lean mass is the biggest open cost^3,2.
Longo leaves diabetes, anorexia, and diseases treated with drugs to a doctor's decision⁵, and the Buchinger clinic rules out pregnancy, anorexia nervosa, and advanced kidney or liver disease, among others⁶.

References

McDougal DH, Darpolor MM, DuVall MA, Sutton EF, Morrison CD, Gadde KM, et al. Glial acetate metabolism is increased following a 72-h fast in metabolically healthy men and correlates with susceptibility to hypoglycemia. Acta Diabetologica. 2018;55(10):1029-1036.Lab or mechanistic studyPubMed 29931424 · DOI
What it found: Six metabolically healthy men aged 18 to 40 stayed 4 days at a research center and fasted 72 hours on water while wearing continuous glucose monitors. The fast lowered blood glucose and caused frequent hypoglycemic episodes, and brain glial acetate metabolism rose from 53.5% to 61.9% (p = 0.005). The glucose, free fatty acid, ketone, and insulin values on the page come from Table 2 and the results text of the full text.
Limits: Six men in one lab fast, designed to study the brain and hypoglycemia rather than health outcomes; the metabolite table is in the full text, not the abstract.
Vendelbo MH, Møller AB, Christensen B, Nellemann B, Clasen BF, Nair KS, et al. Fasting increases human skeletal muscle net phenylalanine release and this is associated with decreased mTOR signaling. PLoS ONE. 2014;9(7):e102031.Lab or mechanistic studyPubMed 25020061 · DOI
What it found: Eight healthy men were studied in the postabsorptive state and after 72 hours of fasting. Forearm net phenylalanine release rose significantly, mTOR phosphorylation fell about 50%, the insulin-stimulated rise in mTOR signaling was blunted, and an autophagy marker rose about 30% while another rose about 10%, making autophagic flux hard to interpret.
Limits: Eight men, one 72-hour fast, and muscle signaling markers rather than measured muscle loss, so it shows the direction of change but not how much muscle a fast costs or whether refeeding restores it.
Ezpeleta M, Cienfuegos S, Lin S, Pavlou V, Gabel K, Varady KA. Efficacy and safety of prolonged water fasting: a narrative review of human trials. Nutrition Reviews. 2024;82(5):664-675.ReviewPubMed 37377031 · DOI
What it found: Narrative review of human trials of prolonged water fasting. Fasting for 5 to 20 days raised ketones and produced 2% to 10% weight loss, about two-thirds of it lean mass; blood pressure consistently fell, lipid results were mixed, fasting glucose, insulin, insulin resistance, and HbA1c fell in people with normal glucose but not in people with diabetes, and benefits were gone 3 to 4 months after refeeding. Reported adverse events included metabolic acidosis, headaches, insomnia, and hunger.
Limits: A narrative (not systematic) review of mostly small, uncontrolled trials of 5 to 20 days, so it does not speak to a single 3-day fast repeated monthly.
Wei M, Brandhorst S, Shelehchi M, Mirzaei H, Cheng CW, Budniak J, et al. Fasting-mimicking diet and markers/risk factors for aging, diabetes, cancer, and cardiovascular disease. Science Translational Medicine. 2017;9(377):eaai8700.Randomized trialPubMed 28202779 · DOI
What it found: Randomized trial of 100 generally healthy US adults comparing an unrestricted diet with a fasting-mimicking diet (low in calories, sugars, and protein, high in unsaturated fats) for 5 consecutive days a month for 3 months; 71 completed three cycles after crossover. Three cycles reduced body weight, trunk and total fat, and blood pressure and lowered IGF-1, with no serious adverse effects, and a post hoc analysis found larger changes in people at risk.
Limits: A diet that supplies some food, not a water-and-broth fast, with surrogate markers over 3 months and a post hoc at-risk analysis; L-Nutra supplied the diet, and two authors, Valter Longo among them, hold equity in it.
Patrick R, Longo V. #27 Valter Longo, Ph.D. on the Fasting-Mimicking Diet & Fasting for Longevity, Cancer & Multiple Sclerosis. FoundMyFitness, October 4, 2016.PodcastListen
What it found: Rhonda Patrick interviews Valter Longo (University of Southern California) about the fasting-mimicking diet, prolonged fasting, and refeeding. In the transcript Longo says the cancer work began in mice, where the effects were very strong, that people usually get a doctor's clearance first, and that for diabetics, anorexic people, and people with diseases who take drugs the doctor has to decide; Patrick says she would not do a multi-day fast without some supervision.
Limits: A 2016 interview read on the show's page (the last quarter of a long page was not read), so it records opinions and hypotheses, and Longo's own papers disclose his equity in L-Nutra, which supplies the fasting-mimicking diet.
Wilhelmi de Toledo F, Grundler F, Bergouignan A, Drinda S, Michalsen A. Safety, health improvement and well-being during a 4 to 21-day fasting period in an observational study including 1422 subjects. PLoS ONE. 2019;14(1):e0209353.Cohort studyPubMed 30601864 · DOI
What it found: One-year observational study of 1,422 people in Buchinger fasting programs of 4 to 21 days at a specialized clinic, eating roughly 200 to 250 kcal a day. Weight, abdominal circumference, and blood pressure fell, ketone bodies rose, 93.2% reported no hunger, 84.4% of 404 people with existing complaints reported improvement, and adverse effects occurred in fewer than 1%.
Limits: Uncontrolled and run by the clinic's own staff and funded through a clinic-linked company (several authors are tied to the clinic), in a supervised residential setting with juice and soup, so it cannot be applied to unsupervised water-and-broth fasting at home.
Zauner C, Schneeweiss B, Kranz A, Madl C, Ratheiser K, Kramer L, et al. Resting energy expenditure in short-term starvation is increased as a result of an increase in serum norepinephrine. The American Journal of Clinical Nutrition. 2000;71(6):1511-1515.Lab or mechanistic studyPubMed 10837292 · DOI
What it found: Eleven healthy, lean volunteers were studied through an 84-hour fast. Resting energy expenditure rose from 3.97 to 4.53 kJ/min by day 3, serum norepinephrine rose from 1,716 to 3,728 pmol/L by day 4, and serum glucose fell from 4.9 to 3.5 mmol/L (all P < 0.05); the authors suggest the glucose drop drives the norepinephrine rise.
Limits: Eleven lean volunteers in one 84-hour lab fast, and the abstract reports no ketone levels.
Cahill GF Jr. Fuel metabolism in starvation. Annual Review of Nutrition. 2006;26:1-22.ReviewPubMed 16848698 · DOI
What it found: A retrospective review by a pioneer of starvation physiology on how insulin and the metabolism of fatty acids, glycerol, glucose, lactate, and pyruvate behave in prolonged starvation; it says D-beta-hydroxybutyrate fuels the brain during starvation and may have therapeutic value when energy supply is reduced, such as stroke, heart insufficiency, and some mitochondrial disorders.
Limits: A personal, narrative review whose abstract gives no data, so I use it for the mechanism, not for numbers.
FoundMyFitness. Topic: Fasting. FoundMyFitness, July 11, 2026.ArticleRead
What it found: Rhonda Patrick's fasting topic page (last updated July 11, 2026): it separates time-restricted eating, alternate-day approaches, prolonged fasting (over 48 hours, typically 48 to 96 hours in research), and the fasting-mimicking diet, says prolonged fasting may be more appropriate under a physician's guidance, and notes that epidemiological studies of fasting cannot show cause and effect.
Limits: A topic page with no named author, and it gives no human outcome numbers for prolonged fasting.
Brandhorst S, Choi IY, Wei M, Cheng CW, Sedrakyan S, Navarrete G, et al. A Periodic Diet that Mimics Fasting Promotes Multi-System Regeneration, Enhanced Cognitive Performance, and Healthspan. Cell Metabolism. 2015;22(1):86-99.Lab or mechanistic studyPubMed 26094889 · DOI
What it found: Mostly yeast and mouse work: periodic fasting-mimicking cycles in middle-aged mice were reported to extend lifespan, reduce visceral fat and cancers, rejuvenate immune cells, and improve cognition in old mice, and a human pilot of three cycles lowered biomarkers for aging, diabetes, heart disease, and cancer without major adverse effects.
Limits: The main findings are in yeast and mice, and the human part is a small pilot of biomarkers, so it is hypothesis-generating for people.
Anderson JC, Mattar SG, Greenway FL, Lindquist RJ. Measuring ketone bodies for the monitoring of pathologic and therapeutic ketosis. Obesity Science & Practice. 2021;7(5):646-656.ReviewPubMed 34631141 · DOI
What it found: Review of how ketones are measured: nutritional ketosis (keto-adaptation) corresponds to blood beta-hydroxybutyrate of at least 0.5 mM, a person eating a balanced diet sits near 0.1 mM, and ketoacidosis starts above 3 mM, with diabetic ketoacidosis ranging from 3 to 20 mM.
Limits: A review of measurement methods; ketosis thresholds vary a little between sources and are conventions, not sharp biological lines.
Masedunskas A, de Ciutiis I, Hein LK, Ge A, Kong YX, Qi M, et al. Investigating the Impact of Glycogen-Depleting Exercise Combined with Prolonged Fasting on Autophagy and Cellular Health in Humans: A Randomised Controlled Crossover Trial. Nutrients, 2024.ArticlePubMed 39770918 · DOI
What it found: A trial protocol, with no results, for a randomized crossover study of a three-day water-only fast with or without glycogen-depleting exercise, measuring autophagy in blood-derived cells.
Limits: A protocol paper with no outcome data (and a participant count that differs between sections), so it shows that autophagy after a 3-day fast in people is still being tested, not that it happens.
da Silva JSV, Seres DS, Sabino K, Adams SC, Berdahl GJ, Citty SW, et al. ASPEN Consensus Recommendations for Refeeding Syndrome. Nutrition in Clinical Practice. 2020;35(2):178-195.Position standPubMed 32115791 · DOI
What it found: ASPEN consensus statement for hospitalized adults and children that defines refeeding syndrome as a 10% to 20% (mild), 20% to 30% (moderate), or over 30% (severe) drop in serum phosphorus, potassium, or magnesium within 5 days of reintroducing calories, with organ dysfunction or thiamin deficiency marking severe cases.
Limits: A consensus definition for hospital patients, meant as a basis for research, so it does not give the risk after a short fast in a healthy adult.
Eli Lilly and Company (FDA-approved prescribing information). ZEPBOUND (tirzepatide) injection, solution; ZEPBOUND KWIKPEN (tirzepatide) injection, solution. DailyMed, National Library of Medicine, August 28, 2026.Regulatory noticeRead
What it found: Current FDA label for tirzepatide, a related GIP and GLP-1 drug (label revised August 2026). Its warnings include severe gastrointestinal adverse reactions (section 5.2), acute kidney injury due to volume depletion (5.3), and hypoglycemia when combined with insulin secretagogues such as sulfonylureas (5.7), where hypoglycemia was 10.3% versus 2.1% without a sulfonylurea.
Limits: The label is for tirzepatide, not retatrutide, which is investigational and adds a glucagon receptor effect, so it shows what a neighboring drug warns about and not what retatrutide's label would say.
Peters AL, Buschur EO, Buse JB, Cohan P, Diner JC, Hirsch IB. Euglycemic Diabetic Ketoacidosis: A Potential Complication of Treatment With Sodium-Glucose Cotransporter 2 Inhibition. Diabetes Care, 2015.ArticlePubMed 26078479 · DOI
What it found: Case series of 13 episodes of ketoacidosis or ketosis in nine people taking SGLT-2 inhibitors (seven with type 1 diabetes, two with type 2); blood glucose was not markedly high, so recognition was slow. The authors advise testing ketones in anyone on these drugs who is nauseated, vomiting, or unwell, and extreme caution in type 1 diabetes.
Limits: A small case series identified incidentally, so it cannot give how often this happens and does not test a keto diet itself.
Patrick R. #040 Dr. Eric Verdin on Ketogenic Diet Longevity, Beta-Hydroxybutyrate & HDAC Inhibitors. FoundMyFitness, December 13, 2017.PodcastListen
What it found: Interview with Eric Verdin, head of the Buck Institute, on how ketogenic diets might affect aging: ketones acting as HDAC inhibitors, PPAR-alpha's role, declining NAD+ with age, keto safety, and intermittent fasting as an alternative.
Limits: A podcast conversation from 2017, not a trial; it discusses mechanisms and hypotheses.