Electrolytes on a 36-hour fast.
Almost every headache, cramp and wave of fatigue people blame on "the fast" is really about sodium. The mechanism is well understood. The specific advice you will read about it mostly is not. Those two things get blended together constantly, so here they are separately.
What is actually established
Well supportedWhen you stop eating, insulin falls. Falling insulin reduces how much sodium your kidneys reabsorb, so you excrete more of it - and potassium with it. This has a name: the natriuresis of fasting. It peaks in roughly the first one to four days, which is exactly the window a 36-hour fast sits in.
A 2025 scoping review in Frontiers in Nutrition catalogued the symptoms that cluster in that window. Across studies of adults starting carbohydrate restriction: muscle cramps in 3-37%, headache in 8-25%, fatigue in 18-25%, dizziness in 15-21%. The pattern is consistent enough that "keto flu" is a real phenomenon with a plausible mechanism.
What is convention, not science
Expert opinionThe numbers you see repeated everywhere - 1-2 g of sodium a day, around 4 g of potassium - trace back to Volek and Phinney. They are reasonable and widely used. They are also not the output of a trial.
The same review that documents the symptom rates is unusually blunt about this. It notes that electrolyte replacement is "heavily encouraged" but has had "little rigorous study and quantification of this requirement," and states plainly that the authors could not identify a single study documenting both keto-induction symptoms and the concomitant changes in serum or urinary sodium and potassium - so even the causal link between the symptoms and the electrolyte loss "needs to be confirmed."
That is worth sitting with. The most repeated practical advice in the entire fasting world rests on a mechanism that makes sense and a dosing recommendation nobody has properly tested.
The one thing that genuinely lands people in hospital
Drinking large volumes of plain water with no sodium during an extended fast can cause hyponatremia. In a chart review of 768 medically supervised water-only fasts, adverse events were mostly mild - but two were serious enough to require hospitalisation, one for dehydration and one for hyponatremia. "Drink lots of water" is the most common fasting advice on the internet and, on its own, it is the most dangerous.
What I do
My experience onlyNot a recommendation. Just what I have settled on after tracking it:
- Salt in water, morning and afternoon. I do not measure it precisely; it is roughly a quarter teaspoon each time.
- Plain black coffee, unchanged from a normal day.
- Magnesium in the evening - mostly because it helps me sleep on fast nights, which is a comfort thing, not a metabolic claim.
- Potassium I get from food on either side of the fast rather than during it.
- If a headache shows up, salt first. It resolves it more often than not, which is weak evidence but it is my evidence.
The one genuinely useful thing tracking gave me: the fasts that went badly were almost always the ones where I forgot the salt on the first morning. That correlation showed up in the log long before I noticed it in the moment.
The fermented-food overlap
A cup of sauerkraut carries roughly 0.9 g of sodium, plus potassium. That is a real contribution - but it lands in your eating window, not during the fast, so it does not solve the intra-fast problem. What it does do is push your weekly sodium higher than you might realise once you add salted food and fasting supplementation on top. If you are managing blood pressure, track the total rather than each source separately.
If you take blood pressure medication, diuretics, insulin, or sulfonylureas, none of the above applies to you without a conversation with your prescriber first. See who should not do this.