⏱️ Fasting & TRE · 14 min read · Part 4 of 7

Prolonged Fasts: 24–72 Hours

The deep end: what a 24–72 hour fast actually does, hour by hour — the ketosis sequence that is real, the autophagy claims that are mostly borrowed from mice, and the electrolyte and re-feeding rules that keep the whole thing from going wrong.

🔎 Evidence Snapshot ★★☆☆☆ Limited — the metabolic sequence is settled old science; the human longevity and autophagy claims rest on animal data

What the evidence supports

The fuel sequence is well documented: glycogen drains, fat oxidation and ketones rise, and the brain shifts partly to ketone fuel (Cahill 2006; Owen 1967).
Supervised multi-day fasts are mostly uneventful in otherwise healthy adults — 1,422 fasts in one observational series (Wilhelmi de Toledo 2019).
Re-feeding risk is well characterized and preventable with a slow, staged return to food.

What remains uncertain

Whether 24–72 h fasts trigger meaningful autophagy in humans — it can't be measured directly, and the marker evidence is almost entirely mouse work.
Any longevity claim in humans: no trial links occasional multi-day fasts to lifespan; the leap from calorie-restricted animals to fasting humans is a leap.
Whether 72 hours beats 24 hours on any measured human endpoint — nobody has shown it.

Evidence last reviewed: August 13, 2026. Conclusions may change as new research is published.

🏔️

the deep end, mapped

Hour by Hour: The Phases

Hours	Phase	What's happening
🍬 0–12	Fed → glycogen	Insulin falls; the body spends the last meal, then liver glycogen takes over
🔥 12–24	Glycogen → fat	Liver glycogen nears empty; fat oxidation rises and ketones appear (~0.3–0.5 mM)
🛢️ 24–48	Ketosis	Ketones ~1–2 mM; the brain shifts partly to ketone fuel; hunger typically blunts
🧬 48–72	Deep ketosis	Ketones ~2–3 mM. Autophagy markers rise in mice; human evidence stays indirect
🏥 72+	Supervised only	Beyond three days is clinical territory — clinics, not kitchens

Ketones by Hour

The switch from sugar to fat, in approximate blood levels (illustrative)

Electrolytes: The Actual Numbers

🧂 Sodium: 3–5 g per day — supervised protocols target this range through broth or salted water; roughly ½ teaspoon of salt (~1.1 g sodium) in water, spread across the day, prevents the day-two headache and wobble.
🍌 Potassium: 2–3 g per day — from potassium salts (lo-salt) or potassium chloride powder in water. Do not megadose: large single doses are a cardiac risk.
⚡ Magnesium: 300–400 mg per day — citrate or glycinate, not oxide; prevents cramps and the wired-tired sleep disruption of deep fasting.
💧 Water: 2.5–3 L per day — more if active. Thirst alone under-reads during fasts; drink on a schedule, not on demand.
⚠️ The cramp rule — cramps, headache, or heart flutter on day 2+: sodium first, then potassium, then magnesium. More plain water alone dilutes the problem; it doesn't fix it.

Re-feeding: How to End It

🥣 First meal small and soft — cooked vegetables, broth, a modest portion of protein; roughly half a normal meal. The gut has been idle; volume is the first risk, not the food itself.
🐟 Second meal normal but plain — easy proteins and starches. The celebratory grease bomb waits for day two.
⏳ Ramp over 24–48 hours — after a 72-hour fast, eat light for the rest of day one and ease into normal portions on day two.
🚨 Re-feeding syndrome, taken seriously — after long fasts, a sudden return to carbohydrates can crash blood phosphate and potassium (Mehanna 2008). Under 72 hours in a well-nourished adult the risk is low; with frailty, alcohol use, or longer fasts, medical supervision is the rule.

Who Shouldn't Even Consider It

🤰 Pregnancy and breastfeeding — no version of this applies.
⚖️ Underweight (BMI under 18.5) or unintentional weight loss — there is nothing in reserve to fast with.
🧠 Any eating-disorder history — prolonged fasting is the disorder's native habitat.
💊 Insulin, sulfonylureas, or other glucose-lowering medications — the hypoglycemia risk is not theoretical.
💊 Diuretics, lithium, digoxin, and any electrolyte-sensitive drug — fasting shifts blood levels.
🫀 Liver, kidney, or cardiac disease; any acute illness — a multi-day fast is a stress test, not a treatment.

Three Days Before: The Taper

The fast starts days before the last meal. Three cheap moves make hour 24 arrive as a normal Tuesday instead of a crisis:

🍬 Day −3: cut the sugar — end the desserts and liquid sugar first. A fast begun from a sugar-heavy week starts with a full glycogen reservoir plus a crash.
🍽️ Days −2 and −1: shrink the meals — one moderate meal less each day teaches the ghrelin schedule that food is not coming on the old clock, and it costs nothing.
🧂 Day −1: start the sodium — 1–2 g of sodium in water the day before softens the day-two deficit, because electrolytes should start the fast topped up, not drained.
🚫 The taper is not the fast — three days of dieting before a 36-hour fast turns the fast into day four of a crash diet. Shrink, don't starve.

Running a 36-Hour Fast

A 36-hour fast — dinner on day one, nothing but the allowed list until breakfast on day three — is the entry point to the deep end: long enough for real ketosis, short enough to stay inside the territory where the safety data is actually decent. Run it by the clock, not by feel:

−12 to 0 h: stage the exit first — a protein-heavy, normal-sized dinner, eaten early enough that the fast starts with the kitchen closed. Measure out the electrolytes and clear the calendar: no first-time fast on a deadline week, before a trip, or beside a hard training block.
0 to 12 h: the boring shift — you spend the last meal, then liver glycogen. Nothing to do except drink on schedule (2.5–3 L a day, not on thirst) and skip the after-dinner graze. 0.5–1 g of sodium in water before bed heads off the day-two headache before it forms.
12 to 24 h: the switch — glycogen nears empty, ketones appear, and hunger usually peaks somewhere in this block. Ride the wave (the temptation-defense page owns the tactics), keep sodium and potassium on the 3–5 g / 2–3 g targets, and keep activity light — a long walk, not a hard session, because blood sugar is at its lowest here.
24 to 30 h: the plateau — ketones around 1–2 mM, appetite often blunts, and the wobble risk is highest. Dizzy or racing on standing is usually volume plus sodium: 500 mL of water with ~1 g of sodium, sit, reassess in 20 minutes. Not better? Eat — the fast is not a hostage negotiation.
30 to 36 h: prepare the landing — write the refeed now (small, soft, half a normal portion), while hunger is honest. The worst refeed decisions happen at hour 36, on hunger plus pride.
36 h and beyond: break it by the refeed rules — small and soft first, plain second, normal across 24–48 hours. Then log how it actually went: hours, doses, wobbles, and how the landing felt. The next fast repeats what worked, not what hurt.

The 36-Hour Arc

Five waypoints, with the wobble zone marked (illustrative)

The Refeed Sequence, Scaled by Fast Length

Fast length	First meal	Second meal	Back to normal
🕛 24 h	Small and easy	Normal	Same day
⏳ 36 h	Half a normal portion, soft	Plain protein and starch	Next day
🏔️ 48–72 h	Broth, cooked vegetables, modest protein	Small and plain	24–48 hours later

The longer the fast, the slower the landing — and under 72 hours in a well-nourished adult, the sequence is comfort and safety, not emergency medicine.

Where the Evidence Lives

This page is the operating manual — the science behind each number has a home in the pillars, and this series references rather than repeats it:

⏱️ The switch itself — the Fasting & Time-Restricted Eating topic owns the metabolic-switch evidence this page's phase table condenses. The pillar owns the chemistry; this page runs the clock.
🧬 The hormesis question — the Fasting as Hormesis topic owns the "is the stress the point?" debate, including why the dose-response curve bends back down. This page operationalizes one dose: 24–72 hours, occasionally.
😰 The cortisol cost — fasting raises cortisol, and the Cortisol topic owns what that does to sleep, mood, and recovery. That is why this page says no first-time fasts on deadline weeks.
🥚 What the refeed should contain — the Protein topic owns the 1.6–2.2 g/kg target the refeed meals are built around; this page only sequences them.
🕗 The lower-risk alternative — the 16/8 Foundations page delivers most of the demonstrated benefit at a fraction of the risk. Read this page as the occasional supplement, not the upgrade.

What to Do When It Goes Wrong

Five ways a prolonged fast goes sideways, and the response that has worked — with the medical line drawn where the evidence draws it:

🚩 Dizzy or lightheaded on standing — the classic hour-24 wobble: blood volume down, sodium down. 500 mL of water with ~1 g of sodium, sit for 20 minutes. Fainting, confusion, or no improvement: break the fast and get medical review — this is the one symptom you do not push through.
🚩 Nausea on day two — usually bile on an empty stomach or too much salt at once. Sip slowly, halve the next electrolyte dose, try plain hot tea and a short walk. Nausea with vomiting or dizziness is a stop sign, not a complication to outlast.
🚩 A hunger wave at hour 18 — normal: ghrelin still pulses on the old meal schedule. Waves pass in 60–90 minutes; water, salted water, and a walk shorten them. If every wave is a cliff, the fast was too abrupt — next time, shrink the last two meals first.
🚩 Can't sleep — fasted sleep is lighter and late-day cortisol sits higher. No caffeine after noon, magnesium glycinate before bed, and expect night one to be rough — which is why the fast never starts the evening before a big day.
🚩 The refeed went wrong — a huge hour-36 meal lands as bloating, cramps, or a rebound sugar slump. Under 72 hours in a well-nourished adult it is usually uncomfortable, not dangerous — but vomiting, diarrhea, or a racing heart after a long fast deserves medical attention, and the next fast gets a planned half-portion first meal.

Questions, Answered Briefly

❓ Is 36 hours enough for autophagy? — No one can measure autophagy in a living human, and the marker evidence is mostly mouse work. What you can say: ketosis, yes; autophagy, unknown.
❓ Should I train during the fast? — Light and daily: walking, easy mobility. Hard training mid-fast costs muscle and raises the wobble risk; put intensity in the fed hours.
❓ Can I do a 36-hour fast every week? — The safety data covers occasional fasts, and nothing shows more frequency is better — most of the demonstrated benefit overlaps with daily 16/8. Treat 24–72 hours as a monthly-or-less tool.
❓ What if I stop early? — Then it was a shorter, successful fast with better data. Stopping at 24 or 30 hours is not failure; pushing through symptoms is the only failure mode on this page.

🏔️ The honest verdict

The animal data is real: calorie-restricted mice and worms live longer, and autophagy machinery exists in us. But no human trial shows that 24–72-hour fasts extend human life, and the demonstrated human benefits — modest weight and metabolic shifts — mostly overlap with what a daily 16/8 and a calorie deficit already deliver. If the deep end calls, go rarely, supervised, and for the experience — not for a longevity claim the data can't yet make.

The Bottom Line

The metabolic sequence is real and old science — glycogen, then fat, then ketones.
Autophagy and longevity claims are animal-derived — the human leap is unproven.
Sodium, potassium, and magnesium are the difference between fine and miserable — track them by number.
End slowly — small, soft, staged — and if you're in any risk group, stay on the shore.

This Page in One Workflow

Clear it — run the "Who Shouldn't Even Consider It" list; if any box is yours, the page ends there.
Stage it — protein-heavy early dinner, electrolytes measured out, calendar light.
Run it — water on schedule, sodium/potassium/magnesium on the numbers, activity light.
Land it — break at the planned hour with a small, soft first meal; ramp across 24–48 hours.
Log it — hours, doses, wobbles, and the refeed's report; the next fast repeats what worked.

The Daily Checklist — On Fast Days

Water on schedule: 2.5–3 L sipped across the day, not chugged on thirst
Sodium 3–5 g across the day, in water or broth — the day-two headache is a sodium debt
Potassium 2–3 g and magnesium 300–400 mg taken, not skipped
Activity light: walking and mobility, no hard training
Symptoms checked honestly — dizziness, heart flutter, or confusion means break and reassess
Caffeine done by noon; magnesium glycinate before bed

The Weekly Checklist — Around Fasts

Fast placed on a light week — never before a deadline, trip, or heavy training block
The "Who Shouldn't" list re-checked before every fast — status changes over a year
Refeed menu written before the fast starts — hour 36 is a bad place to plan
The finished fast logged: hours, doses, wobbles, refeed reaction
Frequency audited — 24–72 hour fasts stay occasional, not weekly