When people talk about "fasting working," they're usually talking about ketosis — though they don't always realize it. The metabolic switch from glucose-and-glycogen fuel to fat-and-ketone fuel is the engine driving most of what people feel and measure during a short fast: the appetite suppression on day two, the mental clarity some people report on day three, the visceral-fat changes over weeks of cycles, the insulin-sensitivity shifts that show up in clamp studies. It is also one of the best-characterized parts of human metabolism — the foundational paper, Cahill 1970, maps the substrate transitions across multi-week starvation in humans with a precision the modern fasting literature still relies on.
This page summarizes what nutritional ketosis is, how short (3–5 day) fasts engage it, and what is distinctive about reaching ketosis on a sardine-only protocol versus a water fast or a long-term ketogenic diet. The applied translation — what ketone numbers to track, when to expect them, what to do if they're not where they should be — is in the ketosis Mechanism Dossier.
What this mechanism is
Ketosis is the metabolic state in which the liver produces ketone bodies — primarily β-hydroxybutyrate (βHB) and acetoacetate — at rates high enough that they become a meaningful systemic fuel. Under normal fed conditions, ketones are present in trace amounts (under 0.2 mmol/L of βHB in blood). "Nutritional ketosis" is conventionally the range from about 0.5 to 3.0 mmol/L of βHB, the level at which brain, muscle, and other tissues begin oxidizing ketones in significant proportion. Diabetic ketoacidosis is a separate, pathological condition reaching 10–25 mmol/L with severe acidosis — it is not what happens during a healthy adult's short fast, and conflating the two is one of the most persistent confusions in popular nutrition writing.
The substrate-level picture is straightforward. The liver stores about 80–120 g of glycogen — perhaps 18–24 hours of fuel at typical resting metabolism, less with exertion. Once liver glycogen is depleted, the liver shifts to gluconeogenesis (producing glucose from amino acids and glycerol) and ketogenesis (producing βHB and acetoacetate from fatty acids delivered by adipose lipolysis). The brain, which cannot oxidize fatty acids, can oxidize ketones — and over the first one to two weeks of fasting, brain ketone uptake can supply 60–70% of cerebral energy demand, sparing the relatively limited gluconeogenic capacity for glucose-obligate tissues. This is the picture mapped meticulously in Cahill's classic paper and confirmed in subsequent isotope-tracer work.
What makes ketones interesting biologically isn't only that they are an emergency fuel. β-hydroxybutyrate also acts as a signaling molecule — it is an HDAC (histone deacetylase) inhibitor, modulates inflammasome activation, and binds receptors that regulate inflammation and oxidative stress. Some of the proposed long-term benefits of ketosis (anti-inflammatory effects, oxidative stress reduction, neuroprotection in animal models) are likely mediated by βHB-as-signal rather than only βHB-as-fuel. Veech 2004 is the canonical review of this signaling biology.
How short fasts engage it
Most healthy adults reach nutritional ketosis (≥ 0.5 mmol/L βHB) somewhere between hour 24 and hour 72 of a fast, depending on:
- Glycogen status going in. A person who has been low-carb for the prior week enters ketosis within 24 hours of starting a fast. Someone coming off a high-carbohydrate diet may take 48–72 hours.
- Activity level during the fast. Light exercise depletes glycogen faster and accelerates ketone production. Sedentary fasting prolongs the transition.
- Body composition. Higher lean mass means more glycogen to burn through. People with very low body fat have less substrate for sustained ketogenesis and may not reach the upper range of nutritional ketosis even on extended fasts.
- Individual variation. βHB at a given fasting hour varies substantially between people — what's "normal" at 60 hours can be 0.6 mmol/L in one healthy adult and 2.4 mmol/L in another.
A 3-day fast typically peaks somewhere in the 1.5–3.5 mmol/L range by the end of day three. A 5-day fast adds another half-day to a day at the upper end of that range, with βHB often plateauing or modestly declining as adipose lipolysis stabilizes. The de Cabo & Mattson 2019 NEJM review frames this as the "metabolic switch" — once you've crossed it repeatedly across cycles, the switch becomes faster and the metabolic adaptations more pronounced. The Mattson 2017 review argues this switching itself, not just the absolute ketone level, is what produces neurological adaptation in animal models.
The most relevant short-fast human data comes from intermittent fasting protocols rather than continuous multi-day fasts. Halberg 2005 ran 20-hour fasts every other day for 15 days in healthy men and saw improved insulin sensitivity (the ketosis-and-insulin connection). Brandhorst & Longo's 2015 fasting-mimicking diet protocol hits ketosis with very-low-calorie eating for 5 days each month — a closer cousin to the Sardine Protocol's structure than a true water fast.
How sardine fasting specifically engages this mechanism
Sardine fasting is, mechanistically, a protein-sparing modified fast (PSMF) with omega-3 enrichment. Unlike a water fast — where the body is sourcing protein from muscle and viscera to supply gluconeogenic substrate — a sardine fast supplies dietary protein at maintenance-adjacent levels (typically 80–120 g/day across 4–6 cans of sardines). This has two consequences for ketosis:
It slows entry into ketosis modestly. The dietary protein triggers a small insulin response and supplies amino acids that fuel hepatic gluconeogenesis directly, reducing the demand for fatty-acid-derived acetyl-CoA flowing into ketone production. In practice, this means an experienced sardine faster typically reaches βHB ≥ 0.5 by the end of day two rather than mid-day two — roughly half a day later than a water fast.
It produces a more sustainable ketosis. Because muscle is being protein-spared, the body is not burning lean mass to fund gluconeogenesis. Adipose lipolysis becomes the dominant fuel source, and the βHB plateau on day 3–5 of a sardine fast often runs in the 1.5–2.5 mmol/L range — enough for metabolic effect, low enough to feel functional. This is the pattern Bistrian and Blackburn aimed for when they originally designed the PSMF protocol in the 1970s, and it's largely the pattern that holds today, though with sardines as the protein vehicle rather than the original liquid formulas.
The omega-3 angle is mechanistically separate from ketosis but co-occurs with it. EPA and DHA reduce inflammatory markers and may modestly improve mitochondrial efficiency at the cellular level. During a fast, when ketones are signaling anti-inflammatory effects through their HDAC inhibition, the sardine-derived omega-3 layer adds a second, additive anti-inflammatory input. The Mechanism Dossier covers this co-engagement in detail.
Where the evidence is strongest, weakest
Strong: Cahill's substrate-level mapping of human starvation physiology is decades-old, well-replicated, and not seriously contested. The signaling biology of βHB (Newman & Verdin 2014, Veech 2004) is well-characterized in vitro and in animals. Short-term human ketogenic-diet outcomes — weight loss, glycemic improvement in T2D — are supported by clinical work like Westman 2008, Volek 2009, and the Virta Health 2-year cohort data in Hallberg 2018.
Moderate: Long-term cardiovascular outcomes on ketogenic diets remain debated. Most data are 12-week to 2-year studies; mortality endpoints don't exist. Athletic-performance effects of ketogenic adaptation are mixed — Phinney's 1983 endurance cyclists study suggested adaptation preserved sub-maximal endurance, but the Burke 2017 race walkers study showed clear performance impairment at race-pace efforts. Whoever cites only one of these studies is making a partial argument.
Weak: Human RCT data specifically on 3–7 day sardine-only protocols is essentially nonexistent. The case for the Sardine Protocol's specific cycle structure leans on PSMF research, fasting-mimicking-diet research, and short-term ketogenic-diet research — adjacent literatures that are individually well-developed but don't combine into a single trial of "the protocol as such."
What this means for your cycle
A short summary lives here on the public page; the detailed application — βHB targets per cycle day, what to do if you're not reaching ketosis, how to interpret numbers across multiple cycles, and how the Sardine Protocol's quarterly 21-day Transformation handles sustained ketosis differently than a monthly 5-day cycle — is in the Inner Circle Mechanism Dossier.
In broad strokes:
- Expect βHB ≥ 0.5 mmol/L by end of day 2 if your prior week was moderately low-carb; later if you came off a high-carb week.
- Expect a peak in the 1.5–2.5 mmol/L range by day 3–5.
- If you don't reach 0.5 mmol/L by mid-day three, something is wrong — see the dossier troubleshooting guide.
- Track βHB at the same time of day each cycle for trend-line comparison; absolute numbers vary across labs and meters.
Open questions
- The human evidence for βHB as a signaling molecule (anti-inflammatory, neuroprotective) is largely extrapolated from animal models and in vitro work. Human dose-response for the signaling effects is not characterized.
- We don't know with confidence how much of the ketosis-attributable benefit in short fasts is the βHB level itself versus the calorie deficit versus the metabolic-switching frequency. Studies that isolate these factors are hard to design and rare.
- Whether repeated short ketogenic exposures produce durable insulin-sensitivity or mitochondrial-efficiency gains beyond the fasting window is genuinely open. The closest evidence — Halberg 2005, Sutton 2018 — suggests yes, but these are small studies.
- Individual variation in ketogenic response is large and not well predicted by baseline characteristics. Some healthy adults plateau at 1.0 mmol/L on day 4 of a fast; others reach 3.5. We don't know why with confidence.