fasting.com

Medical caution

This is general information about fasting, not medical advice. Fasting can affect blood sugar, blood pressure, medication timing, and nutritional status. Talk to a clinician before starting or changing a fasting practice if you are pregnant, under 18, living with diabetes or another chronic condition, taking prescription medication, or have a history of an eating disorder.

What Counts as a Prolonged Fast

Prolonged fasting usually refers to 48 hours or more without food, water only. This extends past the 24-to-72-hour range covered on water fasting, which is the more commonly attempted entry point, and the broader landscape of fasting types covered on what is fasting; this page picks up where those leave off, focused on what changes as a fast continues well past the point most people stop, and on the risks that become more material the longer a fast runs. Trials of prolonged fasting as a treatment for a specific diagnosed disease are a separate area and are not summarized here.

The Stages of a Prolonged Fast

As a fast lengthens, the body moves through a reasonably well understood metabolic sequence, described in the general fasting-physiology literature rather than measured hour by hour in a single trial that followed people across weeks (Cahill, 2006) [1].

Roughly the First Day: Glycogen Depletion

In the first day without food, the body draws down its stored glycogen, the short-term reserve of glucose held in the liver and muscles. A single 24-hour water-only fast significantly increased human growth hormone, red blood cell count, hematocrit, and total and HDL cholesterol, and decreased triglycerides and weight, compared with a day of normal eating, in a randomized crossover trial of 30 apparently healthy volunteers; in the 16 participants who fasted first, most of these changes, including growth hormone and cholesterol, were back to baseline after a following day of normal eating, except weight, with triglycerides also staying suggestively lower (Horne et al., 2013) [2]. A secondary analysis of the same trial found the growth-hormone rise was independent of weight loss and larger in people who started with lower growth-hormone levels, correlating with changes in glucose, hemoglobin, and IGF-1, a mechanism finding rather than a health outcome on its own (Horne et al., 2025) [3].

Roughly Day 2-3: The Shift to Ketone Metabolism

As glycogen runs out and insulin falls, the body shifts from burning glucose to breaking down fat into ketone bodies, with beta-hydroxybutyrate becoming a major fuel for the brain (Cahill, 2006) [1] and sparing muscle protein that would otherwise be broken down for glucose (Palmer and Clegg, 2021) [6]. That protein sparing is partial, not complete: in the two studies in that review that measured body composition (an 8-day and a 10-day fast, one by bioimpedance, which the authors note is unreliable when hydration changes), roughly two-thirds of the weight lost was lean mass (Ezpeleta et al., 2024) [10]. This shift is what makes a multi-day fast physiologically sustainable at all (Rebello et al., 2025) [14]. A 72-hour water-only fast in five healthy adults found a biphasic pattern across several blood markers, cholesterol, an inflammation marker, and thyroid hormones rising or falling by the end of the fast and then moving back toward baseline after 11 days of refeeding, with only minor, reversible change in liver enzymes and electrolytes; insulin and glucose fell without reaching statistical significance in this small, 5-person pilot (Lauc et al., 2026) [4].

Beyond Day 3: Deepening Ketosis

The most detailed day-by-day human data cited here just past this point comes from a 7-day water-only fast in 12 healthy volunteers, which found distinct, body-wide protein changes, flagged by the study's own authors as pointing to both possible benefits and possible harms, appearing only after day three of complete calorie restriction, alongside an average weight loss of 5.7 kilograms (Pietzner et al., 2024) [11]. The kidney also adapts as ketosis deepens: nitrogen loss shifts from urea toward ammonia as the kidney increasingly uses glutamine, and the brain's growing use of ketones reduces the need to draw amino acids from muscle, conserving protein, while adjustments in urine sodium help limit potassium loss, and changes in urine pH help limit uric acid crystals (Palmer and Clegg, 2021) [6]. That same review names ketoacidosis from SGLT2-inhibitor medication, alongside diabetic and alcoholic ketoacidosis, as a related condition sharing the same underlying chemistry, directly relevant to the medication caution below (Palmer and Clegg, 2021) [6]. For how fasting is thought to trigger autophagy, the cell's internal recycling process, see autophagy and fasting; that mechanism is a separate question from the ketone-metabolism shift described here.

Ending the Fast: The Refeeding Transition

The longer a fast runs, the more deliberately it needs to end. See how to break a fast safely for the full refeeding guidance; in short, reintroducing food too quickly after an extended fast can cause a fall in blood phosphorus, potassium, or magnesium, a condition with a specific clinical definition and severity grading (da Silva et al., 2020) [15].

What Human Studies Show

Short Studies (24 Hours to 7 Days)

The individual human studies on fasting length cited here are a handful of small studies spanning 24 hours to 7 days: a randomized crossover trial of a 24-hour fast in 30 people (Horne et al., 2013) [2], a single-group 72-hour pilot in 5 people (Lauc et al., 2026) [4], and a single-group 7-day fast in 12 people (Pietzner et al., 2024) [11]. These are useful for understanding what happens inside the body over the first week, not for a settled claim about health outcomes at 48 hours and beyond; none followed participants for more than 11 days after the fast ended.

Longer, Supervised Fasts (4-21 Days)

The key evidence on what a longer, supervised fast does to weight and metabolic markers is a one-year observational study of 1,422 people doing a supervised, modified fast of 4 to 21 days, with a small daily intake of 200 to 250 kilocalories rather than strictly zero calories: it found significant reductions in weight, waist circumference, and blood pressure, with adverse effects, documented by the clinic's physicians, in fewer than 1% of participants by the authors' own summary (14 recorded events in 1,422 people, including two hospitalisations, one for a heart attack on fasting day 9 in a 75-year-old man with known coronary artery disease; the authors describe the other 12, which included 3 cases of irregular heart rhythm, 3 of low blood sodium, and 2 of low blood sugar, as transitory, and none of them interrupted the fast), while self-reported mild symptoms were more frequent (sleep disturbance in 15%, fatigue in 14%, and hunger in 7% of participants who returned the daily symptom questionnaire) (Wilhelmi de Toledo et al., 2019) [12]. This study had no control group. A chart review of 768 strictly water-only fasting visits under medical supervision found the highest-grade adverse event was severe (grade 3) in 212 visits and life-threatening (grade 4) in 1 visit, with no deaths, and 2 of 768 visits involved a serious adverse event (Finnell et al., 2018) [13]. Diarrhea appears in both studies, recorded in different ways. In the one-year observational study, whose program also gave a laxative at the start of the fast and an enema or mild laxative every second day, 1,311 of the 1,422 participants returned a daily symptom checklist, and 38 of those 1,311 reported diarrhea (about 3%), counted only when the same participant reported it repeatedly during the fast (Wilhelmi de Toledo et al., 2019) [12]. The chart review, coding symptoms from clinical notes across both the fast and the refeeding period, recorded diarrhea in 109 of 768 visits (14.2%), mostly graded mild, and recorded constipation less often (Finnell et al., 2018) [13]. Because the programs, populations, and recording methods differ, the two rates are not directly comparable. A narrative review of human trials of prolonged water-only fasting lasting 5 to 20 days found weight loss of 2% to 10% of body weight; in the two studies in that review that measured body composition (an 8-day and a 10-day fast, one by bioimpedance, which the authors note is unreliable when hydration changes), roughly two-thirds of the weight lost was lean mass and one-third fat mass, which the authors flag as a concern suggesting increased muscle protein breakdown. What happens to that lean mass after the fast ends is not settled: a DXA-based follow-up of 29 adults with overweight or obesity after a median 14-day supervised water-only fast found fat-free mass partly regained by a follow-up visit about 6 weeks after refeeding, with fat loss making up most (67%) of the weight still off (Gabriel et al., 2025) [19], while a 10-day supervised fast with a small daily calorie allowance and daily supervised exercise in 16 healthy men found body weight and lean tissue still below baseline at 3 months while fat mass was no longer significantly different from baseline (Laurens et al., 2021) [20]; see "Where the studies disagree" in the box above for how the studies differ. Blood pressure consistently decreased, lipid effects were inconsistent, and glycemic measures improved in people with normal blood sugar but not in people with type 1 or type 2 diabetes. Reported adverse events included headaches, insomnia, fatigue, dizziness, hunger, and dry mouth, with no serious adverse events, metabolic acidosis, or deaths reported in the trials reviewed, though the review notes more severe events, including metabolic acidosis, documented in earlier reports; the metabolic improvements were no longer seen 3 to 4 months after the fast, even when the weight loss was maintained (Ezpeleta et al., 2024) [10]. A randomized, waitlist-controlled trial of a 5-day modified fast (38 healthy adults aged 20 to 50, 200 to 250 kilocalories a day from vegetable juice and broth, not strictly water-only, done at home after group instruction from certified fasting counselors) points the same way with a stronger design: reductions in BMI and fat mass partly persisted at a 12-week follow-up, lean mass lost during the fast had fully returned, and most cardiometabolic, microbiome, and metabolomic changes measured right after the fast had largely reversed (Kaufhold et al., 2026) [21].

On body composition, a 2026 meta-analysis of 49 studies of water-only fasts of 24 hours or longer, pooling before-and-after changes rather than comparisons with a control group, found fat-free mass fell (Hedges' g=-0.51) along with total body water (g=-0.68), body weight (g=-0.38), BMI (g=-0.42), fat mass (g=-0.25), and waist circumference (g=-0.55); these are standardized effect sizes, not kilograms. Weight loss was larger in fasts longer than 3 days and in participants with obesity, but meta-regression found no significant relationship between fasting duration and the size of the change in BMI, fat mass, or fat-free mass. The authors conclude that the weight lost during a fast includes substantial fat-free mass and body water, not fat alone (Ulupınar et al., 2026) [17].

The lipid finding in that review, effects "inconsistent" from trial to trial, now has a more specific shape. A 2026 meta-analysis of 32 water-only-fasting studies, mostly small before-and-after studies without a control group, found HDL cholesterol fell overall (Hedges' g=-0.233), with no change in fasts of 3 days or less and a clear fall in longer fasts (g=-0.325, 95% CI -0.451 to -0.200). LDL cholesterol rose overall (g=0.489) and in fasts of every length, climbing until roughly day 10 and then levelling off or partly reversing. Total cholesterol rose overall (g=0.343), most in fasts longer than 3 days, and levelled off or eased slightly after about day 5. Triglycerides showed no overall effect (g=-0.039), with trends toward a fall in fasts of 3 days or less and a rise in longer fasts that were not statistically significant. The authors rate the certainty of this evidence low for HDL and very low for the other lipids, report signs of publication bias for LDL and total cholesterol, and describe the duration thresholds as hypothesis-generating rather than clinically actionable (Çamli et al., 2026) [18]. Its finding of no HDL change in fasts of 3 days or less differs from the 24-hour trial above, which found HDL rose (Horne et al., 2013) [2].

Two Case Reports at the Extreme End (40 and 44 Days)

Two case reports cited here describe what happened at the far end of fasting length; they are case reports, not trials. A 57-year-old woman admitted to hospital after a 40-day water-only fast had low blood sodium on admission, then developed low potassium, phosphate, and magnesium during refeeding (Brett and Nesbit, 2013) [16]. A case report on a 44-day voluntary fast that produced a 25% body-weight loss found that, during early refeeding, despite a cautious start with a low-sodium, low-calorie liquid feed, the person developed marked low blood phosphate, fluid dilution, and mild swelling; vitamins B1, B12, and B6 were depleted; free fatty acids, ketone bodies, and zinc were abnormally high; liver function became abnormal over the first week; and the appetite hormones leptin and ghrelin were very low at the start of refeeding, with hunger rising along with a transient rise in two other hormones, orexin and resistin, as refeeding continued (Korbonits et al., 2007) [5]. Each is one person, not a generalizable finding; together, these are the two documented extreme-length cases cited here, showing what refeeding danger can look like after a very long fast.

Risks That Rise With Length

Almost all of the safety data on fasts of several days or longer come from medically supervised settings: a residential medical facility in the chart review of 768 water-only fasting visits (Finnell et al., 2018) [13], and a specialised fasting clinic in the 1,422-person study (Wilhelmi de Toledo et al., 2019) [12]. Their low rates of serious events describe supervised fasting. They cannot be read as the risk of a multi-day fast done alone at home. One home-based trial cited here was a 5-day modified fast in 38 screened healthy adults aged 20 to 50 taking no prescription medication, with a fasting counselor reachable throughout; 83% reported fasting-related adverse events, all mild and transient, and none stopped the fast (Kaufhold et al., 2026) [21]. The other, a 5-day at-home modified fast in 32 screened healthy adults, with adverse events recorded by study physicians, reports only transient events: one woman had brief nausea and vomiting on day 2, and three participants who worked together, one of them in the control group, reported symptoms such as dizziness and nausea on day 2, which the authors say may have had an environmental cause; all continued (Grundler et al., 2026) [24]. Both describe short, screened, supported fasts, not an unsupervised water-only fast of the length this page covers, and no study cited here measures that risk. In the chart review of 768 water-only fasting visits, presyncope (feeling close to fainting) was recorded in 217 visits (28.3%), all graded moderate, and the authors describe it as known to occur in response to fasting (Finnell et al., 2018) [13]. In the 1,422-person modified-fast study, medical staff documented dizziness in 2 of the 1,422 enrolled (0.14%), and 28 of the 1,311 who returned the daily symptom questionnaire reported vertigo (about 2%), counted only when the same person reported it repeatedly; the authors report that average blood pressure fell during the fast without falling below the normal range, and that serious hypotensive complications were not reported (Wilhelmi de Toledo et al., 2019) [12]. In the home-based 5-day modified-fast trial, "circulatory dysregulation", a term the paper does not define, was listed for 10 of the 38 participants (26.3%), the most common entry in its supplementary table of fasting-related adverse events (Kaufhold et al., 2026) [21]. None of the three papers reports whether these symptoms happened on standing up. As a general caution rather than a finding from these studies, a drop in blood pressure on standing up (orthostatic hypotension) can cause dizziness or fainting. Anyone considering a fast longer than 48 hours should ask a clinician whether it should be done under medical supervision.

Refeeding Syndrome

Refeeding syndrome is a fall in blood phosphorus, potassium, or magnesium of 10% or more within five days of reintroducing calories, with severity graded by how far levels drop (da Silva et al., 2020) [15]. The two cases above show directly what that can look like at the extreme end (Brett and Nesbit, 2013 [16]; Korbonits et al., 2007) [5]. For the full risk-factor and incidence picture, see how to break a fast safely; this page states the risk plainly rather than re-deriving that detail.

Electrolyte Problems

Plain water has no electrolytes, and the longer a fast runs, the more the body's sodium, potassium, and magnesium balance can shift as glycogen and its associated water stores deplete. See electrolytes during a fast for the mechanism behind this. In a chart review of 768 medically supervised water-only fasting visits, both serious adverse events were fluid or electrolyte problems: a life-threatening low-sodium event on fasting day 9 and severe dehydration on fasting day 3, each in a different person, both of whom needed hospital care and recovered; the review does not report what caused either event (Finnell et al., 2018) [13]. In the 1,422-person supervised-fasting study, low blood sodium was among the adverse events documented by the clinic's medical staff in 3 cases (0.21% of the 1,422 enrolled); the paper's discussion of blood results separately reports six mild cases during the fast, with the lowest sodium reading at 127 mmol/L, all non-serious and corrected with sodium chloride during the supervised program, and it does not explain why the two counts differ (Wilhelmi de Toledo et al., 2019) [12]. A separate case report describes a woman admitted to hospital after a 40-day water-only fast with low sodium, who then developed low potassium, phosphate, and magnesium during refeeding (Brett and Nesbit, 2013) [16]. This is a general caution rather than a dosing guide, and nothing here is a substitute for a clinician's guidance on electrolyte replacement during a prolonged fast.

Medication Interactions (Diabetes Medication, SGLT2 Inhibitors)

SGLT2 inhibitors, a class of medicines used for type 2 diabetes and heart failure, can cause euglycemic ketoacidosis, a dangerous build-up of ketones without high blood sugar. A 2025 review describes this as a well-recognized adverse event in people with diabetes taking these medicines, and summarises six published cases in people without diabetes taking one for heart failure; in five of the six, the common risk factor was eating much less than usual because of illness, fasting, or surgery (Garg et al., 2025) [7]. The six cases show that the risk exists in people without diabetes too; they cannot say how often it happens. Anyone taking an SGLT2 inhibitor should talk to the prescribing clinician before any fast, and should not stop or change the medicine without that advice. More broadly, anyone on any diabetes or blood-pressure medication should not start any fast without that conversation first, and the same general caution applies to any other prescription medication, including blood thinners, since no study cited here tests a specific fasting-medication interaction beyond the SGLT2 class.

Gallstones (Evidence From Very-Low-Calorie Diets)

Gallstones are a recognized risk of rapid weight loss on very-low-calorie diets, which still provide some food each day and are not the same as a water-only fast. A 1993 review of human studies estimated that between 10% and 25% of obese men and women may develop gallstones within a few months of starting a very-low-calorie diet, and that perhaps one third of those will develop symptoms; people with the highest body mass index before weight loss and those who lose weight most rapidly appeared to be at greatest risk (Everhart, 1993) [22]. A 16-week trial of a liquid diet of 520 kilocalories a day, in people with a body mass index of 38 or more and a normal gallbladder ultrasound at the start, was testing a gallstone-prevention medicine; in the group given a placebo instead, gallstones were found on ultrasound in 28% (95% CI 22% to 35%) (Shiffman et al., 1995) [23]. The two supervised fasting studies cited on this page do not settle the question for fasting. The published adverse-event tables of the chart review of 768 water-only fasting visits name no gallbladder or biliary event, though they list 1,356 of its 5,961 recorded events only as "other" (Finnell et al., 2018) [13], and the 1,422-person study of a supervised, modified fast of 4 to 21 days lists no gallbladder or biliary event among its symptoms and adverse effects (Wilhelmi de Toledo et al., 2019) [12]. Neither study used ultrasound to look for gallstones, so neither can show whether gallstones that cause no symptoms formed. Other human studies have looked at the gallbladder during different kinds of fasting, such as the daytime fast of Ramadan, long overnight gaps between meals, and several days without food after abdominal surgery; those are different situations and are not summarized here. No study cited here has measured gallstone risk during or after a water-only fast, so the diet findings above are a reason for caution, not a rate for water fasting, and this page does not turn them into one.

Who Should Talk to a Clinician Before a Prolonged Fast

Talk to a clinician before any fast, of any length, if you take any regular prescription medication, manage diabetes or blood pressure with medication, have kidney disease or heart failure, take a diuretic, are pregnant, are under 18, have a history of an eating disorder, a mood disorder such as depression or bipolar disorder, or a psychotic disorder, or are receiving cancer treatment. On psychiatric conditions specifically: a 2025 narrative review of the psychological and psychiatric effects of fasting (87 studies, 39 of them in humans) reports that randomized trials found few adverse events and no signal of severe psychiatric harm, while observational studies more often reported symptom exacerbation in at-risk groups, including mood destabilization, worsened anxiety, and rare psychotic or manic decompensations in people the review describes as vulnerable; the review names active eating disorders, unstable mood or psychotic disorders, and treatment with medicines that need precise dosing or timing, such as lithium, as conditions in which fasting "should be avoided or undertaken only with close medical supervision" (Bonaccorsi and Romeo, 2025) [27]. On pregnancy specifically: no study cited here tests prolonged water fasting in pregnancy directly. The closest available evidence is a meta-analysis of observational studies of daytime fasting in pregnancy, mostly during Ramadan, with no randomized trials; it found a small reduction in birth weight (about 94 g) without a higher rate of low birth weight (Giorno et al., 2025) [8]. That is a different kind of fasting and cannot show what a fast of 48 hours or longer does in pregnancy; this gap in the evidence is itself a reason for caution, not a basis for a specific claim about prolonged fasting in pregnancy.

How This Differs From a Shorter Water Fast

Water fasting covers the more commonly attempted 24-to-72-hour range: what it is, whether it is reasonably safe to try, and a first-pass safety list. This page covers a different question: what changes in the body as a fast continues well past that range, and the risks, refeeding syndrome, electrolyte shifts, and medication interactions, that become more material the longer a fast runs. The two pages deliberately overlap in the 48-to-72-hour band, since the stages described above genuinely span that window too; see "Where the studies disagree" in the box above for why that overlap is intentional. For related but shorter fasting patterns, see alternate-day fasting and hunger during intermittent fasting.

Authors and Funding of the Cited Studies

  • Wilhelmi de Toledo et al., 2019 [12]: the lead author, FWT, is a member of the Directory Board of the Buchinger Wilhelmi Clinic (BWC), where the study was performed, and, as managing director of Amplius GmbH, executes the scientific leadership at BWC; Amplius GmbH "conceives, coordinates and develops fasting research on behalf of BWC." Co-author FG is currently employed, and co-author SD was formerly employed, at BWC. Co-author AM is a consultant at BWC and receives financial compensation for that role from Amplius GmbH.
  • Finnell et al., 2018 [13]: "ACG is the director of the TrueNorth Health Center, president of the board of the TrueNorth Health Foundation, and member of the board of the National Health Association. He does not receive financial compensation for these roles. BCS and TRM are paid consultants for the TrueNorth Health Foundation, which is funded in part by the TrueNorth Health Center. JSF received funds from the National Health Association grant." Funding: "This research was supported in part with a grant from the National Health Association. Grant number not provided. Additional support was provided by the TrueNorth Health Foundation. The funding bodies played no role in the design of the study or the collection, analysis, and interpretation of data."
  • Horne et al., 2025 [3]: one author, BH, discloses membership on the advisory board of Unleash Health and past consulting for Pfizer on risk scores, both "outside this work."
  • Garg et al., 2025 [7], Giorno et al., 2025 [8], Pietzner et al., 2024 [11], and Rebello et al., 2025 [14]: each paper states the authors declare no competing interests.
  • Çamli et al., 2026 [18]: funded by the Scientific Research Projects Coordination Unit of Istanbul Yeni Yuzyil University. Disclosure: "The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest."
  • Gabriel et al., 2025 [19]: the paper states, "The authors received no specific funding for this work." Conflicts of interest, quoted from the paper: "D.M.G. consults for Metabite Inc. A.C.G. is owner of TrueNorth Health Center and President of the Board of the TrueNorth Health Foundation. All other authors declare no conflicts of interest." D.M.G. is Goldman; A.C.G. is Goldhamer. The paper's author list gives the TrueNorth Health Foundation as the affiliation of Gabriel, Ncube, and Myers.
  • Laurens et al., 2021 [20]: funded by the CNRS and Amplius GmbH (Überlingen, Germany), with an additional grant from the Laboratoire Therascience (Monaco) for hormonal and metabolic assays; the paper states no additional external funding was received. Conflicts of interest, quoted from the paper: Françoise Wilhelmi de Toledo "is managing director of Amplius GmbH, a company that coordinates the scientific documentation for the Buchinger Wilhelmi Clinics and is co-worker of the Buchinger Wilhelmi Clinic," and Franziska Grundler "is co-worker at the Buchinger Wilhelmi Clinic." The paper states the other ten authors declare no conflict of interest.
  • Kaufhold et al., 2026 [21]: funding, quoted in part from the paper: "Open Access funding enabled and organized by Projekt DEAL. This work was funded by intramural funds of the Experimental and Clinical Research Center, a cooperation between the Max Delbrück Center for Molecular Medicine in the Helmholtz Association and Charité - Universitätsmedizin Berlin; and external funds from the Karl and Veronica Carstens foundation (KVC 0/120/2021)." The paper also lists support for individual authors from the German Federal Ministry of Research, Technology and Space (BMFTR), the European Union's Horizon programme, the Berlin Center for Translational Vascular Biomedicine, the DZG Innovation Fund, the German Research Foundation (DFG), and the DZHK. Competing interests, quoted from the paper: "NS has received speaker honoraria from Teva GmbH, Alexion Pharma Germany GmbH and Roche Pharma AG. RM is employed by the Buchinger Wilhelmi Development & Holding GmbH. The other authors declare no competing interests." NS is Siebert; RM is Mesnage.
  • Everhart, 1993 [22], and Shiffman et al., 1995 [23]: the full text of neither paper was available to us, so their funding and conflict-of-interest statements could not be read.
  • Grundler et al., 2026 [24]: funding, quoted from the paper: "The study was funded by Buchinger Wilhelmi Development and Holding GmbH." The paper also lists academic fellowships for individual authors from the LUMC, the Health + Life Science Alliance Heidelberg Mannheim, and EMBO. Competing interests, quoted from the paper: "Authors FG, RM, AH and FWT are employees of the Buchinger Wilhelmi Development and Holding GmbH. Authors SS, SH, and RG are employees of Lifespin GmbH. The remaining authors declare no competing interests." FG is Grundler, the first author; RM is Mesnage, the corresponding author; AH is Holley; FWT is Wilhelmi de Toledo; SS, SH, and RG are Strathmeyer, Heelemann, and Geyer.
  • Solianik et al., 2026 [25]: funding, quoted from the paper: "This work was supported by the Research Council of Lithuania (Grant Number S-MIP-23-84). Research Council of Lithuania had no role in the design, analysis or writing of this article." Competing interests, as printed: "The authors declare no competing interests."
  • Bouck et al., 2025 [26]: funding, quoted from the paper: "TDG was supported by a Killam Postdoctoral Fellowship. HI was supported by a Michael Smith Health Research BC Research Trainee Award." Conflict of interest statement, as printed: "None."
  • Bonaccorsi and Romeo, 2025 [27]: the paper's statements are printed in the singular although two authors are listed. Conflicts of interest, as printed: "The author declares no conflicts of interest." Funding, as printed: "This research received no external funding. The article-processing charge (APC) was self-funded by the author."
  • Disclosures are listed where the full text of the paper was available to us; a study not listed here is not a statement that its authors have no ties.

Safety and Cautions

This is general information, not medical advice. The direct human evidence on fasting length specifically is small, short-term-heavy at the controlled-trial end, and largely uncontrolled at the longer end; a systematic review of the broader fasting literature found only three randomized controlled trials in humans through January 2015, though those trials did show improvements in weight and other risk-related outcomes, and concluded that more research is needed before fasting can be recommended as a health intervention (Horne et al., 2015) [9]. Talk to a clinician before any fast, of any length, if you are pregnant, under 18, have a history of an eating disorder, a mood disorder such as depression or bipolar disorder, or a psychotic disorder, have kidney disease or heart failure, take a diuretic, take any regular prescription medication, manage diabetes or blood pressure with medication, or are receiving cancer treatment. SGLT2-inhibitor medication carries a specific, documented ketoacidosis risk when combined with fasting (Garg et al., 2025) [7]. Stop and seek care for any severe or concerning symptom, and refeed gradually; see how to break a fast safely. For the broader pooled evidence on fasting generally, see what the research shows about fasting.

What the evidence actually supports

Direct human evidence on fasting length specifically is thin. The individual studies cited here are a handful of small studies spanning 24 hours to 7 days: a randomized crossover trial of a 24-hour fast in 30 people (Horne et al., 2013) [2], a single-group 72-hour pilot in 5 people (Lauc et al., 2026) [4], and a single-group 7-day fast in 12 people (Pietzner et al., 2024) [11]; only the first of these is controlled. Beyond that, the evidence is a large observational, uncontrolled study of a modified, not strictly zero-calorie, supervised fast up to 21 days (Wilhelmi de Toledo et al., 2019) [12], and a chart review of medically supervised water-only fasts (Finnell et al., 2018) [13]. The physiology that explains why longer fasts are survivable at all, the shift from glycogen to ketone-body metabolism, is well established as general science (Cahill, 2006 [1]; Rebello et al., 2025 [14]; Palmer and Clegg, 2021 [6]), but the stages described on this page are built from that mechanism, not from a single trial that followed people hour by hour across weeks. Two case reports cited here, at the extreme end of fasting length (40 and 44 days), show what can go wrong during refeeding (Brett and Nesbit, 2013 [16]; Korbonits et al., 2007 [5]); each is one person, not a generalizable finding. Two 2026 meta-analyses from overlapping author groups pool many small water-only-fasting studies to test fasting length directly, one on body composition (49 studies) and one on blood lipids (32 studies); both pool before-and-after changes rather than comparisons with a control group, and the lipid review rates its own evidence low to very low certainty (Ulupınar et al., 2026 [17]; Çamli et al., 2026 [18]). Three further studies extend the follow-up window past refeeding: two small uncontrolled trials that tracked body composition for weeks to months after a multi-day fast, one strictly water-only and one with a small daily calorie allowance, disagree with each other on what happens to lean mass (Gabriel et al., 2025 [19]; Laurens et al., 2021 [20]), and a randomized, waitlist-controlled trial of a 5-day modified fast followed 38 healthy adults for 12 weeks, finding that reductions in BMI and fat mass partly persisted while most cardiometabolic, microbiome, and metabolomic changes measured right after the fast had largely reversed (Kaufhold et al., 2026) [21]. A second randomized trial of a 5-day at-home modified fast, run by the same Buchinger Wilhelmi clinic group that produced the uncontrolled 4-to-21-day cohort cited above, compared 32 healthy adults on the program with 32 who kept their usual diet: weight fell more in the fasting group (p<0.001), blood pressure was not significantly lower at the end of the fast itself but was after the 4-day food reintroduction, and the gut-microbiome and blood-metabolite changes seen after the fast no longer differed from the control group one month later, with blood pressure back at pre-intervention levels; the authors' own conclusion is that the program "is safe and transiently improves cardiometabolic health" (Grundler et al., 2026) [24]. A single-group study of a 6-day fast in 14 non-obese, physically inactive men found maximal voluntary contraction, jump performance, and fine motor skills preserved through the fast despite increased subjective fatigue on days 4 and 6, with vigor up and fatigue and depressive symptoms down after 7 days back on a habitual diet (Solianik et al., 2026) [25]. A randomized crossover trial in 15 healthy young adults accustomed to high-intensity exercise, each completing a 60-hour fast with and without a 3-hour cycling bout (one dropped out of the exercise condition), found both conditions lowered the inflammatory cytokine TNF-alpha and raised the anti-inflammatory cytokine IL-10, with the IL-10 rise appearing to be driven by the exercise condition, and total white blood cell count unchanged, which the authors say suggests immune function is not compromised by a fast of that length (Bouck et al., 2025) [26]. A gallstone caution further down draws on studies of very-low-calorie diets, not fasting (Everhart, 1993 [22]; Shiffman et al., 1995 [23]); the two supervised fasting studies above name no gallstone events, but neither looked for gallstones with ultrasound.

Where the studies disagree

This page deliberately overlaps with water fasting in the 48-to-72-hour range; both pages draw on the same narrative review (Ezpeleta et al., 2024) [10] and the same supervised-fasting studies (Wilhelmi de Toledo et al., 2019 [12]; Finnell et al., 2018 [13]), because the stages framework on this page genuinely needs to describe that window too, not because the overlap was missed. The SGLT2-inhibitor caution below rests on a 2025 narrative review that describes euglycemic ketoacidosis as a well-recognized adverse event in people with diabetes taking these medicines, and summarises six case reports in people without diabetes taking one for heart failure (Garg et al., 2025) [7]; it names the risk but gives no rate. The closest available evidence on pregnancy is a meta-analysis of observational studies of daytime fasting in pregnancy, mostly during Ramadan, with no randomized trials; it found a small reduction in birth weight (about 94 g) without a higher rate of low birth weight (Giorno et al., 2025) [8]. That is a different kind of fasting and cannot show what a fast of 48 hours or longer does in pregnancy. Across the evidence as a whole, a systematic review that searched through January 2015 found only three randomized controlled trials of fasting in humans, though those trials did show improvements in weight and other risk-related outcomes, and concluded that substantially more research is needed before fasting can be recommended as a health intervention (Horne et al., 2015) [9]; the risk side of this page's content, refeeding syndrome, electrolyte shifts, and medication interactions, is better documented than the stages narrative and should carry the most editorial weight. A further disagreement concerns what happens to lean mass after the fast ends. One study followed 29 adults with overweight or obesity (median age 62, 23 of them women; 29 of 40 enrolled completed the follow-up scan) through a median 14-day supervised water-only fast, a 6-day refeed onto a whole-plant-food diet, and a follow-up visit a median 45 days after that, measuring body composition by DXA: fat-free mass was partly regained after the fast, and by the follow-up visit fat loss made up most (67%) of the weight still off (Gabriel et al., 2025) [19]. Another followed 16 healthy men (mean BMI 26) through a 10-day supervised fast that allowed 200 to 250 kilocalories a day from juice, broth, and honey, combined with up to 3 hours a day of supervised low-intensity exercise, with a 3-month follow-up measured by bio-impedance in 12 of them: body weight and lean soft tissue were still significantly below baseline, while fat mass was no longer significantly different from baseline (Laurens et al., 2021) [20]. Neither study had a control group, and both papers attribute part of the lean mass lost during the fast itself to water rather than tissue. A third, shorter and controlled trial of a 5-day modified fast in 38 healthy adults found lean mass lost during the fast fully regained by 12 weeks (Kaufhold et al., 2026) [21]. The studies differ in population, protocol, fast length, refeeding diet, measurement method, and follow-up length; this page does not pick a winner between them, and none of them is read here as what happens to all fasters.

References

  1. Cahill GF Jr. (2006). Fuel metabolism in starvation. Annual Review of Nutrition. PMID 16848698 doi:10.1146/annurev.nutr.26.061505.111258 Finding: limited evidence
  2. Horne BD, Muhlestein JB, Lappé DL, May HT, Carlquist JF, Galenko O, Brunisholz KD, Anderson JL (2013). Randomized cross-over trial of short-term water-only fasting: metabolic and cardiovascular consequences. Nutrition, Metabolism and Cardiovascular Diseases. PMID 23220077 doi:10.1016/j.numecd.2012.09.007 Finding: supports
  3. Horne BD, Anderson JL, May HT, Bair TL, Le VT, Iverson L, Knowlton KU, Muhlestein JB (2025). Weight loss-independent changes in human growth hormone during water-only fasting: a secondary evaluation of a randomized controlled trial. Frontiers in Endocrinology. PMID 39991046 doi:10.3389/fendo.2024.1401780 Finding: limited evidence
  4. Lauc G, Brlek P, Bulić L, Šimunić Briški N, Šimunović J, Duvnjak Orešković I, Butumović L, Marjanović D, Klarić D, Petrović A, Tesla Frcko J, Primorac D (2026). Systemic metabolic, hormonal, and glycomic remodeling during a 72-hour fast in healthy adults: a pilot study. Croatian Medical Journal. PMID 42286908 doi:10.3325/cmj.2026.67.226 Finding: limited evidence
  5. Korbonits M, Blaine D, Elia M, Powell-Tuck J (2007). Metabolic and hormonal changes during the refeeding period of prolonged fasting. European Journal of Endocrinology. PMID 17656593 doi:10.1530/EJE-06-0740 Finding: limited evidence
  6. Palmer BF, Clegg DJ (2021). Starvation Ketosis and the Kidney. American Journal of Nephrology. PMID 34350876 doi:10.1159/000517305 Finding: limited evidence
  7. Garg R, Sood N, Bansal O, Hoskote A (2025). Euglycemic Ketoacidosis Associated with SGLT-2 Inhibitors in Non-diabetic Patients - A Narrative Review. Journal of General Internal Medicine. PMID 39354257 doi:10.1007/s11606-024-09073-2 Finding: supports
  8. Giorno A, De Simone C, Lopez G, Pisaturo ML, Niccolini L, Guida M, Sarno L, Schettini SCA (2025). Intermittent Fasting During Pregnancy and Neonatal Birth Weight: A Systematic Review and Meta-Analysis. Nutrients. PMID 41305597 doi:10.3390/nu17223546 Finding: limited evidence
  9. Horne BD, Muhlestein JB, Anderson JL (2015). Health effects of intermittent fasting: hormesis or harm? A systematic review. American Journal of Clinical Nutrition. PMID 26135345 doi:10.3945/ajcn.115.109553 Finding: limited evidence
  10. Ezpeleta M, Cienfuegos S, Lin S, Pavlou V, Gabel K, Varady KA (2024). Efficacy and safety of prolonged water fasting, a narrative review of human trials. Nutrition Reviews. PMID 37377031 doi:10.1093/nutrit/nuad081 Finding: supports
  11. Pietzner M, Uluvar B, Kolnes KJ, Jeppesen PB, Frivold SV, Skattebo Ø, Johansen EI, Skålhegg BS, Wojtaszewski JFP, Kolnes AJ, Yeo GSH, O'Rahilly S, Jensen J, Langenberg C (2024). Systemic proteome adaptions to 7-day complete caloric restriction in humans. Nature Metabolism. PMID 38429390 doi:10.1038/s42255-024-01008-9 Finding: limited evidence
  12. Wilhelmi de Toledo F, Grundler F, Bergouignan A, Drinda S, Michalsen A (2019). Safety, health improvement and well-being during a 4 to 21-day fasting period in an observational study including 1422 subjects. PLoS One. PMID 30601864 doi:10.1371/journal.pone.0209353 Finding: supports
  13. Finnell JS, Saul BC, Goldhamer AC, Myers TR (2018). Is fasting safe, a chart review of adverse events during medically supervised, water-only fasting. BMC Complementary and Alternative Medicine. PMID 29458369 doi:10.1186/s12906-018-2136-6 Finding: supports
  14. Rebello CJ, Zhang D, Anderson JC, Bowman RF, Peeke PM, Greenway FL (2025). From starvation to time-restricted eating, a review of fasting physiology. International Journal of Obesity. PMID 39369112 doi:10.1038/s41366-024-01641-0 Finding: supports
  15. da Silva JSV, Seres DS, Sabino K, Adams SC, Berdahl GJ, Citty SW, Cober MP, Evans DC, Greaves JR, Gura KM, Michalski A, Plogsted S, Sacks GS, Tucker AM, Worthington P, Walker RN, Ayers P (2020). ASPEN Consensus Recommendations for Refeeding Syndrome. Nutrition in Clinical Practice. PMID 32115791 doi:10.1002/ncp.10474 Finding: supports
  16. Brett AS, Nesbit RM (2013). A 40-day water-only fast by a Pentecostal woman: clinical and religious observations. The American Journal of the Medical Sciences. PMID 23232813 doi:10.1097/MAJ.0b013e3182760349 Finding: limited evidence
  17. Ulupınar S, Terzi M, Çamli A, Asan S, Şeran B, İnce İ, Gençoğlu C, Çabuk S, Yağan E, Özbay S (2026). Duration-Dependent Changes in Body Composition During Prolonged Fasting: A Systematic Review and Meta-Analysis. Nutrition Reviews. PMID 42440276 doi:10.1093/nutrit/nuag092 Finding: supports
  18. Çamli A, Ülker İ, Terzi M, Yağan E, Özbay S, Ulupınar S (2026). Duration-dependent effects of water-only fasting on blood lipids: a systematic review, meta-analysis, and threshold meta-regression. Frontiers in Nutrition. PMID 41994097 doi:10.3389/fnut.2026.1772246 Finding: supports
  19. Gabriel S, Ncube M, Goldman DM, Scharf E, Goldhamer AC, Myers TR (2025). Prolonged Water-Only Fasting Followed by a Whole-Plant-Food Diet Promotes Fat-Free Mass Recovery and Continued Fat Mass Loss in Adults With Overweight or Obesity. Obesity Science & Practice. PMID 40765844 doi:10.1002/osp4.70086 Finding: limited evidence
  20. Laurens C, Grundler F, Damiot A, Chery I, Le Maho AL, Zahariev A, Le Maho Y, Bergouignan A, Gauquelin-Koch G, Simon C, Blanc S, Wilhelmi de Toledo F (2021). Is muscle and protein loss relevant in long-term fasting in healthy men? A prospective trial on physiological adaptations. Journal of Cachexia, Sarcopenia and Muscle. PMID 34668663 doi:10.1002/jcsm.12766 Finding: limited evidence
  21. Kaufhold G, Bartolomaeus TUP, Schütte K, Schütte T, Kamboj S, Löber U, Rahn G, McParland V, Braun L, Markó L, et al. (2026). Machine learning identifies microbiome and clinical predictors of sustained weight loss following prolonged fasting. Genome Medicine. PMID 42754908 doi:10.1186/s13073-026-01765-0 Finding: supports
  22. Everhart JE (1993). Contributions of obesity and weight loss to gallstone disease. Annals of Internal Medicine. PMID 8214980 doi:10.7326/0003-4819-119-10-199311150-00010 Finding: limited evidence
  23. Shiffman ML, Kaplan GD, Brinkman-Kaplan V, Vickers FF (1995). Prophylaxis against gallstone formation with ursodeoxycholic acid in patients participating in a very-low-calorie diet program. Annals of Internal Medicine. PMID 7755224 doi:10.7326/0003-4819-122-12-199506150-00002 Finding: limited evidence
  24. Grundler F, Ducarmon QR, Holley A, Knufinke M, Strathmeyer S, Heelemann S, Geyer R, Martínez-Téllez B, MacArthur MR, Zeller G, Wilhelmi de Toledo F, Mesnage R (2026). Health benefits of a five-day at-home modified fasting program: a randomised controlled trial. Genome Medicine. PMID 42226305 doi:10.1186/s13073-026-01681-3 Finding: supports
  25. Solianik R, Eimantas N, Daugė G, Brazaitis M (2026). The effect of 6-day fasting on physical performance and neuromuscular control in healthy men. Scientific Reports. PMID 42323358 doi:10.1038/s41598-026-57312-9 Finding: supports
  26. Bouck T, Monteleone J, Duffy J, Ainslie PN, Little JP, Thomas KN, Gibbons TD, Islam H (2025). Changes in plasma cytokines following a 60-h fast are not influenced by the addition of exercise despite elevated ketones in healthy young adults. Physiological Reports. PMID 40129260 doi:10.14814/phy2.70294 Finding: supports
  27. Bonaccorsi V, Romeo VM (2025). Psychological and Psychiatric Consequences of Prolonged Fasting: Neurobiological, Clinical, and Therapeutic Perspectives. Nutrients. PMID 41515178 doi:10.3390/nu18010060 Finding: limited evidence