Fasting is one of the most-marketed and most-misunderstood interventions in human optimization. The popular framing collapses several distinct biological events into a single "fasting works" claim. In reality, fasts of 12, 24, 48, and 72 hours each invoke different physiology. Time-restricted eating (which is mostly a calorie-reduction tool) is mechanistically distinct from a multi-day water fast (which produces a deeper metabolic switch). This page lays out what each duration does, what the trial evidence actually shows, and where the rodent literature outpaces the human data.
What does fasting do to your body?
Within 4 to 6 hours of the last meal, blood glucose returns toward fasting baseline (around 90 mg/dL in healthy adults) and serum insulin falls. The liver shifts from glycogen synthesis to glycogenolysis. Glucagon rises about 2-fold over the next 12 hours, signaling further hepatic glucose output.
By 12 to 18 hours of food abstention, hepatic glycogen stores (around 100 g in a fed adult) approach depletion. Mobilization of free fatty acids accelerates, and the liver begins producing ketone bodies, mainly beta-hydroxybutyrate. Serum ketone levels rise from a fed baseline near 0.1 mmol/L to 0.3 to 0.6 mmol/L by 18 hours. Nutritional ketosis (defined as 0.5 to 3.0 mmol/L) is typically reached around 24 to 36 hours in lean adults, faster in those with prior carbohydrate restriction.
By 48 to 72 hours, the body is running primarily on free fatty acids and ketones. Brain glucose dependence falls from about 100% in the fed state to roughly 40 to 60% by day 3, with ketones supplying the rest. Resting energy expenditure stays close to baseline through the first 72 hours and only starts to drop measurably (around 5 to 10%) after day 4 to 5 in extended fasts. The "starvation mode" myth that REE collapses inside 24 hours is not consistent with calorimetry data.
Growth hormone rises substantially: a 24-hour fast produces GH peaks 5 to 10 times baseline, an adaptive signal that helps preserve lean mass during nutrient absence. This is a real mechanism that does not require pharmacology to invoke.
mTOR and AMPK: the energy-sensing axis
Two protein kinase systems sense cellular fuel state and decide whether the cell builds or recycles. mTOR (mechanistic target of rapamycin) is activated by amino acids (especially leucine), insulin, and energy abundance. It pushes protein synthesis, ribosome biogenesis, and growth. AMPK is activated by rising AMP/ATP ratio (energy depletion). It pushes catabolism, mitochondrial biogenesis, and autophagy.
In the fed state, mTOR is dominant and AMPK is suppressed. Fasting flips this. By 24 hours of food abstention, mTOR signaling in skeletal muscle and liver drops 40 to 60% versus fed controls, and AMPK phosphorylation rises 2 to 4-fold. The ratio matters more than either absolute level. Rapamycin produces a similar mTOR suppression pharmacologically without the AMPK signal, which is why fasting and rapamycin are often discussed as functionally analogous interventions even though the mechanisms diverge.
The downstream effect of low-mTOR / high-AMPK is a shift in cellular bookkeeping toward catabolism: protein synthesis slows, mitochondrial biogenesis upregulates, and the autophagy machinery activates.
Does fasting trigger autophagy in humans?
Autophagy is the cellular recycling pathway that degrades damaged proteins and organelles via lysosomal fusion. Mizushima and Komatsu 2011 reviewed the molecular machinery and the place of autophagy in tissue homeostasis ( Mizushima & Komatsu 2011 ). In rodents, the evidence that fasting upregulates autophagy is robust. Alirezaei 2010 fed mice 24-hour and 48-hour fasts and biopsied multiple tissues, finding profound increases in neuronal autophagy markers ( Alirezaei et al. 2010 ). Liver, muscle, and heart show similar patterns at similar durations.
The human story is thinner. Direct biopsy evidence in humans is small-n. Jamart 2012 took muscle biopsies from 11 ultraendurance runners after a 24-hour event and showed elevated LC3-II (a key autophagy marker) and reduced mTOR signaling, consistent with autophagy induction. Most other human "autophagy" claims rely on indirect blood markers (LC3 in serum, p62, beclin-1) of unclear specificity.
Klionsky 2021 issued the consensus methods guidelines for measuring autophagy and was direct about the gap: most non-biopsy assays are not specific, and a single elevated marker does not prove flux is increased ( Klionsky et al. 2021 ). Static markers can rise either because autophagy is increasing OR because the lysosomal step is blocked. Without flux assays, the strong claims about how many hours of fasting "turn autophagy on" are extrapolations.
The honest synthesis: rodent data strongly supports fasting-induced autophagy at 24 to 48 hours. Human muscle biopsy data is consistent but sparse. The claim that "16 hours of fasting maximizes autophagy" or "X hours flips the switch" is not supported by direct human flux data.
Does intermittent fasting work for weight loss?
The dominant fasting protocol in popular use is time-restricted eating (TRE), typically 14:10 or 16:8. Cienfuegos 2020 (Cell Metabolism, n=58) compared 4-hour and 6-hour feeding windows to control over 8 weeks ( Cienfuegos et al. 2020, n=58 ). Both TRE arms lost about 3% of body weight with comparable insulin sensitivity gains. Lowe 2020 (JAMA Internal Medicine, n=116) compared 16:8 TRE to a 3-meal control with no other timing instruction over 12 weeks ( Lowe et al. 2020, n=116 ). Primary weight-loss endpoint was null. The reconciliation between the two trials hinges on what the control group ate: passive controls are easier to beat than actively-structured 3-meal controls.
Sutton 2018 ran a more carefully controlled 5-week eTRE trial in 8 men with prediabetes using a 6-hour early-window protocol (8 a.m. to 2 p.m.) ( Sutton et al. 2018, n=8 ). The trial held calories constant via supervised feeding. Insulin sensitivity rose, blood pressure fell, and oxidative stress markers improved without weight loss. This is the cleanest evidence to date that the timing itself, not just calorie reduction, can move metabolic markers in a metabolically vulnerable population. The trial is small and the design exhausting to scale.
The synthesis: TRE produces meaningful weight loss in the 1 to 4% range over 8 to 12 weeks when adherence is high, and the bulk of the effect is calorie reduction. Independent metabolic gains beyond calories are real but small in healthy adults and clearer in metabolically impaired ones.
Alternate-day and prolonged fasting
Trepanowski 2017 (JAMA Internal Medicine, n=100) compared alternate-day fasting (25% kcal on fast days) to continuous 25% calorie restriction over 52 weeks ( Trepanowski et al. 2017, n=100 ). Weight loss was comparable at 6%. Dropout was higher in the ADF arm (38% versus 29%), and the ADF group had a slightly worse LDL trajectory. The trial confirms two things: ADF is not metabolically superior to matched calorie restriction in healthy adults, and adherence is the limiting factor.
Prolonged fasting (3 to 5 day water-only or fasting-mimicking diets) has been studied by the Longo lab. Wei 2017 (Science Translational Medicine, n=100) ran a 3-month fasting-mimicking diet RCT: 5 days every month of about 750 kcal, otherwise normal eating ( Wei et al. 2017, n=100 ). Outcomes: 5.7 lb weight loss, BMI dropped 1.0 unit, IGF-1 reduced, fasting glucose lowered. The signal is real and the protocol is more tolerable than water-only multi-day fasts. The longevity extrapolation from rodent FMD lifespan extension to humans is speculative.
Lean mass and muscle protection during fasting
The 5-fold to 10-fold rise in growth hormone during a 24 hour fast is part of the body's lean-mass-protection program. GH stimulates lipolysis, conserves glucose for the brain, and limits proteolysis. The system works to a point. Past about 72 hours, especially in older adults with lower baseline lean mass, the muscle-sparing margin gets thin. Trials of repeated 24 hour fasts in healthy adults under age 50 show roughly preserved fat-free mass when protein at refeeding is adequate (1.6 to 2.0 g/kg/day) and resistance training continues 2 to 3 sessions per week. In older adults or those starting at lower lean mass, the same protocol can lose 0.5 to 1.5 kg of lean mass over 8 to 12 weeks of frequent fasting.
The practical implications:
- Hit total weekly protein at or above 1.6 g/kg/day, even if intra-day distribution is uneven because of fasting windows.
- Keep resistance training in the schedule on at least 2 sessions per week.
- Break extended fasts with about 30 to 40 g of high-leucine protein to reactivate muscle protein synthesis effectively.
- Creatine 5 g/day continues during fast periods (it is not a caloric substrate; clearance is slow).
What about ketosis vs fasting?
A ketogenic diet (under 50 g carbs/day) can produce nutritional ketosis (0.5 to 3.0 mmol/L beta-hydroxybutyrate) without caloric restriction. The metabolic readouts overlap with fasting on some measures (ketones, lower insulin) and diverge on others (mTOR is not necessarily suppressed; lean-mass dynamics depend on protein intake). The autophagy signal in keto-fed rodents is smaller than in fasted rodents at comparable durations, suggesting that low-insulin alone is not as autophagy-permissive as low-insulin plus low-amino-acid plus low-energy.
For a person targeting ketones (say, for migraine prophylaxis or epilepsy), keto delivers without the calorie deficit. For a person targeting the broader low-mTOR / high-AMPK state that fasting produces, keto is a partial substitute and not a full one.
What fasting is NOT
Three claims survive in the popular literature without strong support:
- "Fasting accelerates fat loss beyond what calorie reduction alone produces." Multiple matched-calorie trials show no fasting-specific advantage in healthy adults.
- "Fasting protects lean mass." GH rises and helps somewhat. But sustained protein deficit during ADF or extended fasting still costs lean mass, especially in older adults. The protective effect is real but partial.
- "16 hours = autophagy on." No human flux data supports a sharp threshold. Autophagy is graded, multi-input (leucine, insulin, energy charge), and not yet quantitatively mapped in humans across fast durations.
Operational read
Fasting is a useful tool when it makes calorie control easier, when it fits a person's schedule, or when prolonged fasting is being used deliberately for a metabolic reset. It is not magic. It does not uniquely accelerate fat loss vs matched-calorie eating. Its independent metabolic benefits are clearer in prediabetic and obese populations than in metabolically healthy adults. The autophagy story is real in rodents and underdetermined in humans.
For practical context on choosing a protocol, see Intermittent Fasting Compared.
