Dry Fasting — The Record

The record · multi-day

The form the internet sells, and the arithmetic it never shows.

No controlled human trial of absolute dry fasting beyond 24 hours was located across 57 queries in 10 languages. What exists is physiology and case reporting, and both are here.

The arithmetic that settles the camel claim

The most repeated justification for multi-day dry fasting is that burning fat produces water, so the body supplies its own. The chemistry is real. The magnitude is not.

QuantityMeasured valueSource read
Metabolic water produced from fat oxidationapproximately 250 mL per day, extrapolated at about 2,500 kcal/dayInstitute of Medicine 2005, Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate, chapter 4. No DOI or PMID located for this chapter; NAP book ID 10925. Not derived
Respiratory water loss, sedentary adult250 to 350 mL per day, rising to 500 to 600 mL in active peopleSame chapter
Insensible loss by diffusion through skinapproximately 450 mL per daySame chapter
Minimum obligatory urine outputapproximately 500 mL per daySame chapter

Those three obligatory losses total roughly 1,200 mL a day at the low end, against roughly 250 mL produced. Metabolic water does not close that gap, it narrows it. The deficit compounds every day, and the compounding is the whole danger. A separate Russian-language source puts the yield at 105 to 107 g of water per 100 g of fat, which is the same order and changes nothing about the conclusion.

The camel is not a counterexample. What lets a camel do this is a set of adaptations humans do not have, and the camel’s hump fat is explicitly not its main water source in the primary literature. The figure of a 40 kg hump yielding 43 litres is real arithmetic about fat, and it is irrelevant to a human being.

What the case literature records when this is done anyway

In the absence of trials, what exists is case reporting, and it is not reassuring. One documented continuous dry fast produced multiorgan failure:

MeasuredValue
Serum sodium187 mmol/L
Creatinine14.5 mg/dL
Creatine kinaseabove 33,000 U/L
Organ failureoliguric acute kidney injury attributed to rhabdomyolysis and volume depletion; respiratory failure with pulmonary edema requiring intubation and VV-ECMO; encephalopathy; vasopressor-requiring compromise
Treatmentcontinuous renal replacement therapy, gradual sodium correction

For scale, a healthy serum sodium sits around 135 to 145 mmol/L. In a separate emergency-department cohort of 256 patients presenting with sodium above 160 mmol/L, mortality predictors included sodium above 166 mmol/L and a slow correction rate. The full harms register, with the populations who must not attempt this at all, is on the limits page.

The ketone reading that is not measuring what people think

The strongest single piece of evidence offered for dry fasting is usually a urine ketone strip reading at the top of its scale. That reading is real and it is being misread.

Urine ketone strips measure acetoacetate, with a sensitivity of about 50 mg/L, and acetone at about 500 mg/L. They cannot measure beta-hydroxybutyrate at all, and beta-hydroxybutyrate is the dominant ketone in deep ketosis. A strip reading 160 mg/dL and marked LARGE is an acetoacetate concentration in urine, concentrated further by the very dehydration the fast produces. It is not a measurement of ketosis depth.

When tonicity was actually controlled in a human trial, the result went the other way. Berneis 1999, a randomized crossover restricting fluid for 17 hours, found beta-hydroxybutyrate did not rise significantly. In the one human study where tonicity moved lipolysis and ketogenesis, it moved them in the direction of hydration, not dehydration. And the most dehydrated state in clinical medicine carries the least ketosis.

For comparison, in ordinary water-permitted fasting beta-hydroxybutyrate rises from about 0.091 mmol/L at baseline to 2.233 mmol/L at 72 hours (Nuttall 2020), and from 0.08 to 2.21 mmol/L in a second series (Jørgensen 2015). Roughly 0.5 mmol/L by 24 hours, 1 to 2 by 36, and 2 to 3 by 48 to 72. Withholding water is not what produces those numbers.

The two records this page summarizes

This page states the conclusions. The full research behind them is published whole, with every source, every query and every disconfirming finding:


dryfasting.com has been published since 2014. This evidence record was compiled on 22 November 2026. Every figure on this page states the source it was read from. Where a source carries no DOI or PMID, the row says so rather than supplying one. Nothing is sold on this site.