The short version
- Five days is a real clinical threshold. NICE: "Any patient with negligible food intake for more than five days is at risk" of refeeding syndrome.
- Every reassuring safety number for long fasting came from a residential clinic with daily bloodwork.
- Buchinger's 250 calories of juice and soup is doing electrolyte work, not calorie work. A pure water fast is the harder choice, not the purer one.
- Dry fasting rests on one study of ten people. And the documented harm of a bad refeed is electrolyte, not gut.
Not medical advice
This is a report on what the published research and the people who run fasting protocols actually say. We don't run a clinic, we haven't supervised anyone's fast, and nothing here is medical advice from us. Fasting interacts with real medical conditions and with common prescriptions, and the medication section below is the part to take literally rather than skim. If you take anything for blood sugar, blood pressure, or your heart, talk to the person who prescribes it before you go days without eating.
Five days is where the guidelines change their language
NICE, quoted in a 2008 BMJ review: "Any patient with negligible food intake for more than five days is at risk" of refeeding syndrome. ASPEN's moderate-risk band starts at "none or negligible oral intake for 5-6 days," and its risk table names "Prolonged fasting (eg, individuals on hunger strikes, anorexia nervosa)" outright. Two days is hunger and a headache, which is why our 48-hour guide reads differently.
And the reassuring numbers (adverse effects in 0.7% of 1,422 Buchinger fasters, zero deaths in 768 water-only visits at TrueNorth) came from residential clinics with daily labs and staff who end a fast when potassium drops. Not a kitchen table. Whether to fast at all is a separate argument, and we made it.
Five weeks of prep for five days of water
Daniel Pompa wrote the popular version: a chiropractor by training who sells health programs rather than practicing clinically, no medical degree. The prep is his best material. Beyond Fasting runs seven weeks, five of preparation, week six the fast, week seven breaking it. "Fasting is like a marathon: you need to prepare for it." The goal is fat adaptation, built by cycling keto, fasting and feast days inside a week: 5-1-1, then 4-2-1, then 3-2-2. His slogan: "don't eat less, eat less often." (His two sites define 4-2-1 differently, which tells you how firm the numbers are.)
Name the conflict: the binders, "detox" products and molecular hydrogen he recommends during a fast are sold by his own companies, no clinical evidence supports them here, and autophagy recycles damaged parts inside your cells rather than liberating toxins into your gut for a supplement to catch.
The clinics prep faster, and neither skips bloodwork. TrueNorth: two days minimum of raw and steamed fruit and vegetables, no coffee, alcohol, nicotine, grains, dairy, meat, added sugar, oil or salt. Buchinger: one 600 calorie vegetarian relief day, preceded by a history, a physical, CBC, electrolytes, liver and kidney panels, uric acid, basal TSH and an ECG if indicated. Taper your caffeine too, because withdrawal headache hit 55.5% of caffeine consumers on Ramadan day one, and expect insomnia, the second most common complaint in supervised fasting at 33.5% of TrueNorth visits, clustered early.
What the five days actually do
Ketones land around 4 to 5.6 mmol/L by day five, a 10-day study peaking at 5.616 there, and glucose bottoms out near 3.5 mmol/L on day four. The early weight isn't fat: Kerndt's review puts the first week at 0.9 kg a day, "primarily due to negative sodium balance," and that 10-day study measured 1.20 kg a day across the first three.
Sodium is the one to understand. On water only with no supplementation, serum sodium fell 2.88% by day six and below the reference floor by day nine. In the Buchinger cohort, where people get 200 to 250 calories of juice and vegetable soup, sodium barely moved and six of 1,422 got mild hyponatremia. That soup is doing electrolyte work, not calorie work. Uric acid roughly doubles too, 5.9 to 12.5 mg/dL by day seven in a 1976 study, as urate clearance fell from 5.5 to 1.8 mL/min. Gout is a stop sign.
Pompa's arc is rough for three days, day four "max autophagy," day five "something magical happens" with stem cells and growth hormone. Take the subjective half: the clinics see the same shape, and the 10-day study found a curve rather than a slide, peak depression-dejection on day three and peak anxiety on day six. The mechanisms are his framing, not physiology. There is no human autophagy clock, the two human muscle-biopsy studies that exist barely moved the markers, and the stem cell work is mice plus preliminary chemotherapy data. Growth hormone he gets right: a 1992 study found a five-fold rise in 24-hour GH production over two days.
Water, salt and standing up slowly
The salt mechanics are the same at five days as at two, and our 48-hour guide works through them: you leak about 4 g of salt a day whether you eat any or not, the two clinics disagree about water by more than double, and drinking to thirst beats drinking to a number. What changes over five days is how far it goes. Hyponatremia was the only life-threatening event in TrueNorth's 768 visits, presyncope showed up in 28.3% of them, and their patients don't operate heavy machinery. Stand up slowly. Think about driving. And no published protocol specifies electrolyte doses for a healthy person fasting unsupervised, so any milligram figure, ours or Pompa's, is a judgment call.
A seven-day fasting study found VO2peak down 13% and peak power down 16%, maximal strength unchanged, glycogen roughly halved. Its DXA scan showed 4.6 kg of "lean mass" gone, mostly water and glycogen, and the unchanged strength is the proof. Nobody loses 4.6 kg of muscle in a week. Walk, stretch, stop lifting heavy, which is where both clinics and Pompa land.
Dry fasting rests on one study of ten people
Pompa separates soft dry fasting (no food or water by mouth, showering allowed) from hard (no water contact at all). His claim: "you actually get a three-to-one autophagy with a dry fast," meaning "one day of dry is like three days of water." He describes two days without water himself, his son dry fasting almost four days, and a client going 12 days without food or water.
The entire peer-reviewed literature on continuous multi-day dry fasting is one study: ten people, five days. Everything else indexed is intermittent religious fasting of 11 to 20 hours with overnight rehydration, which says nothing about four continuous days. No human study establishes 3x autophagy, and none has measured autophagy in a dry-fasting human at all.
What that study found: cortisol up 495%, renin up 315%, uric acid up 103%, hematocrit up 11%. Hemoconcentration plus a near six-fold stress response, not a clean bill of health. It publishes no absolute sodium or osmolality values, so anyone quoting you those made them up. And endocrinologists who withhold water diagnostically, with monitoring and IV access, stop at 3% body weight loss. This one recorded about 10%.
Then the case report. A 59-year-old man, BMI 24, no risk factors, took no water for the first five days of a fast and developed a proximal deep vein thrombosis running from the peroneal veins to the common femoral vein, with residual thrombus still there at follow-up. His doctors wrote: "The public needs to be aware of the potential life-threatening complications associated with important dehydration in the setting of medically unsupervised fasting."
Punch the other way too. Acute kidney injury was lower in Ramadan fasters than in propensity-matched non-fasters, 13.4% versus 20.4%, and serum osmolality actually fell across 19 days of daytime Bahá'í dry fasting. Neither transfers: those people rehydrate overnight, and the sickest opt out.
No clinical body has a position on dry fasting. The 2024 international consensus in Cell Metabolism defined the term, found it among the most contested, and said the process "did not aim at evaluating therapeutic effectiveness or safety." Therapeutic fasting medicine has never been dry: Buchinger mandates a minimum 2.5 L of fluid a day. We're not writing you a dry fasting protocol, because writing one would mean inventing the numbers.
Breaking it is the part with the documented harm
The two clinics disagree about what to eat and for how long, and neither has evidence for its version.
| Protocol | First food | The ramp | How long |
|---|---|---|---|
| Buchinger | A ripe apple, raw or cooked, eaten slowly. Potato soup in the evening. | Roughly 800, 1,000, 1,200 and 1,600 calories | About four days, whether the fast ran 5 days or 20 |
| TrueNorth | Juice and vegetable broth | Five phases: juice, raw produce, steamed, grains, then unrestricted whole plant food | At least half the length of the fast |
| Pompa | Berries, soft avocado, steamed or blended vegetables | 500 to 800 calories on day six, 800 to 1,000 on day seven | Two days, then no meat or dense protein for several more |
Zero randomized trials have compared first foods after a prolonged fast. Both protocols are tradition, carefully written down. Pompa's reason for the slow ramp is protecting the stem cells he thinks day five produced; the clinical reason is refeeding syndrome. Right answer, different reason, which is worth saying because it means the advice is safe to relay. Bone broth, the internet's default fast-breaker, is in neither clinic protocol, and Pompa contradicts himself on it. His dedicated fast-breaking article never mentions broth, and neither does its day-by-day food chart. His Pompa Program piece on water fasting names "bone broth, soups, and smoothies" among the best options for breaking one, and he's said on his own podcast that bone stock is "one of the things we use to break the fast with." Note too that his best-known bone broth episode is an affiliate placement for a broth company. The one real paper on it concluded it's "unlikely that bone broth can provide a consistently reliable source of key amino acids."
Refeeding syndrome, plainly: starvation drains phosphate, potassium and magnesium out of your cells while blood levels still look normal, then carbohydrate spikes insulin, insulin drives all three into the cells at once, and the process burns through thiamine.
Two cases make it real. A 30-year-old man ended a 44-day fast and was refed carefully, slow ramp, thiamine cover, and his phosphorus still fell from 1.0 to 0.46 mmol/L on the evening of day one and needed IV replacement. And a 2025 case of Wernicke's encephalopathy after a 40-day water-only fast: phosphate 0.25, potassium 2.8, sodium completely normal. Feeling fine and having normal-looking numbers is not a safety signal.
The counterweight matters as much. Refeeding problems were rare in the supervised cohorts and gastrointestinal disaster isn't documented: eight hunger strikers refed after 43 days had "only one episode of diarrhea" between them, and only 20% of adverse events in TrueNorth's 2024 trial happened during refeeding. The documented harm is electrolyte, not gut.
Which kills the folklore. Enzymes don't switch off; the only human measurement found changes at day ten, still low-normal, lipase unchanged. The microbiome doesn't need repopulating, since alpha diversity was unchanged across a 10-day fast and self-corrected in three months. The human gut tolerates disuse far better than the rodent gut. A big meal feels terrible because the stomach empties slower and the glucose and insulin response is exaggerated, both measured.
Training after has no controlled evidence at all. Glycogen was halved, so resynthesis is the limiter, and clinic practice is gradual reintroduction once you're eating solid food. Convention plus physiology, and we'd rather say so.
The people who should skip it
Buchinger's published absolute contraindications: cachexia, anorexia nervosa and eating disorders, uncontrolled hyperthyroidism, advanced cerebrovascular insufficiency or dementia, advanced liver or kidney insufficiency, pregnancy and nursing. Their relative list, a specialist rather than a flat no, adds type 1 diabetes, unstable coronary disease, psychotic disorders, ulcers, addictions and cancer. Add gout.
Older adults lose two defenses at once. A 1984 NEJM study found healthy elderly men were less thirsty and drank less after 24 hours of water deprivation, despite larger rises in plasma osmolality and sodium than young men. The aging kidney also concentrates urine roughly 20% less well.
Eating disorders deserve more than a checkbox. A five-year prospective study of 496 adolescent girls found fasting predicted the onset of recurrent binge eating and bulimia more strongly and more consistently than dietary restraint did. If that's your history, this isn't your article.
The prescriptions that change the arithmetic
Metformin's own FDA label says the drug "should be temporarily discontinued while patients have restricted food and fluid intake." That's the label, not a commentator. The canagliflozin label says to withhold at least three days before procedures involving prolonged fasting, because SGLT2 inhibitors can produce ketoacidosis while your blood sugar reads normal. Lithium is handled by the kidney the way sodium is, about 80% reabsorbed in the proximal tubule, so volume depletion pushes levels up, and its label tells patients to maintain a normal diet including salt. Warfarin is the counterintuitive one: the replicated signal puts the risk window after the fast, with supratherapeutic INR risk running 69% higher post-Ramadan.
We're not handing you a stop list, and there's a reason. The NHS board that popularized sick-day medication rules says the evidence such advice reduces net harm is "very weak," and names real harms from stopping the wrong drug. Take this section to whoever writes your prescriptions, weeks before you start.
Bottom line
Five days is doable for a healthy adult with an empty medicine cabinet, and the honest version is unglamorous: weeks of prep, a blood panel, salt, a lot of sitting down, and a refeed that takes longer than the fast. Take Pompa's structure and his slow ramp, skip the dry fasting and the supplements he sells. If anything on your prescription list appears above, this starts with a phone call.