The Question Everyone Asks, and the Honest Answer
“How long do I have to fast before autophagy kicks in?” is the single most common fasting question I get, and the internet has answered it with confident numbers: twelve hours, sixteen, twenty-four, seventy-two. Almost all of those numbers come from rodent studies, and a mouse is not a scaled-down human. A mouse burns through its liver glycogen in a few hours and is in deep ketosis by the next morning. We take a day or more to make the same switch, and the cellular recycling program that fasting is supposed to trigger lags behind it.
So the honest answer has two parts. First, the human evidence for autophagy from short daily fasts does not yet exist. Second, the human evidence for multi-day protocols does, and it is more encouraging than most sceptics expect.
What Autophagy Is, in One Paragraph
Autophagy is the process by which a cell tags worn-out proteins and damaged mitochondria, wraps them in a membrane, and digests them for parts. It runs at a low level all the time and ramps up when nutrients are scarce, because a starving cell recycles before it builds. Declining autophagy is one of the recognised hallmarks of aging, which is why fasting, exercise and a handful of drugs that switch it on attract so much longevity interest.1 The catch is that measuring it in a living person is hard. Until recently, every human claim rested on proxies such as ketones or the growth signal mTOR, not on autophagy itself.
The Timing Table: Protocol by Protocol
This is how the evidence actually lines up as of 2026. The middle column describes the metabolic state; the right column is the part that matters, direct human measurement of autophagy.
| Protocol | Metabolic state in humans | Direct human autophagy evidence |
|---|---|---|
| 12-hour overnight fast | Insulin falling; glycogen still largely intact | None. Mechanistically implausible as a strong trigger. |
| 16:8 time-restricted eating | Modest ketone rise by morning; insulin and appetite improve over weeks | None measured. Benefits are metabolic, not (yet) autophagic. |
| 24-hour fast | Liver glycogen mostly depleted; ketones climbing; the “metabolic switch” begins | Limited. Small blood-cell marker studies, no flux measurement. |
| 48–72 hour water fast | Sustained ketosis; mTOR and IGF-1 suppressed; growth hormone up | Suggestive. Autophagy-related protein changes reported in immune cells in small studies. |
| 5-day fasting-mimicking diet | Fasted physiology at 700–1,100 kcal/day of a designed plant-based kit | Strongest. A 2025 trial measured autophagic flux directly; it roughly doubled by day five to six. |
The Study That Changed the Answer
In 2025 a Cedars-Sinai group published the first human trial to measure autophagic flux directly during a fasting intervention.2 Thirty adults were assigned to a five-day fasting-mimicking diet, a second fasting-mimicking formulation, or a control diet. Instead of proxies, the investigators isolated white blood cells at several visits and measured LC3-II turnover with a chloroquine assay, which is the accepted way to show that the recycling machinery is not just switched on but actually processing cargo.
Flux rose about 100% by the third study visit, at the end of the five days, and it was the only statistically significant change in the trial. Fasting glucose, insulin resistance and body weight improved alongside it, as they do in every fasting-mimicking study. The result was in blood cells, not liver or muscle, and the trial was small and industry-funded though academically run. Those are real limits. It is still the only human evidence we have that a defined fasting protocol does what the mouse work predicted. I wrote about the trial in more detail when it came out: Human proof that fasting activates autophagy.
Why Sixteen Hours Probably Is Not Enough
Time-restricted eating is a good tool. Restricting meals to an eight- or ten-hour window lowers fasting insulin, trims a few pounds, and makes people notice what they eat. I recommend it often. But its benefits track with the insulin story, not the autophagy story. Human glycogen stores last roughly 18 to 24 hours; the metabolic switch to fat and ketones, which is the precondition for a meaningful autophagy signal, is usually not complete until well past that point.3 A 16-hour fast ends about where the interesting biology begins.
That does not make 16:8 useless. It makes it the wrong tool if autophagy is the specific outcome you want, and the right tool for most of what people actually need, which is better glucose control and less late-night eating.
Water Fasting vs. Fasting-Mimicking: Which to Choose
A three- to five-day water fast produces a deeper fasted state than a fasting-mimicking diet, and a few small studies have found autophagy-related changes in immune cells over that span.1 It also produces more muscle loss, more electrolyte risk, more headaches, and a much higher abandonment rate on day two. A fasting-mimicking diet was designed to keep the physiology while removing most of those costs: it feeds you enough plant-based, low-protein, low-sugar food that the body still reads it as fasting. It is the protocol with the direct human data, and it is the one I use with patients.
A reasonable cadence for a metabolically healthy adult is one five-day cycle every three months. More often may be appropriate for specific goals, but this is not a weekly practice, and it is not for everyone.
Who Should Not Do Multi-Day Fasts
Anyone with a history of an eating disorder, anyone underweight, anyone pregnant or breastfeeding, and anyone with type 1 diabetes should not attempt a multi-day protocol. People on insulin, sulfonylureas, SGLT2 inhibitors, GLP-1 medications or diuretics need medication adjustments before starting, because hypoglycaemia and electrolyte shifts are real risks. If you take prescription medications of any kind, this is a conversation to have with a physician first, not after.
The Practical Read
Use a daily eating window for metabolic health. Use a supervised five-day fasting-mimicking cycle a few times a year if autophagy and the longevity biology behind it are the goal. Skip the seventy-two-hour water fasts unless someone is monitoring you. And treat any specific hour count you read online as a mouse number until a human trial says otherwise.
Next step
If you are weighing a daily eating window against a supervised multi-day protocol, the right next step is a fit conversation—not another hour count from the internet.
Book a free 30-minute discovery call with Luke Sorrell, MD FACP. He will go through your history, meds, and goals, and tell you honestly whether time-restricted eating, a fasting-mimicking cycle, or something else fits—and whether telemedicine care is available in your state.
Book a free 30-minute discovery call
Already have recent labs? Try the free Lab Decoder to compare glucose, insulin, lipids, and more with the optimal ranges Dr. Sorrell uses. It runs in your browser; nothing is sent or saved. Not a diagnosis—bring questions to the call.
Frequently asked questions
Is 16 hours of fasting enough for autophagy?
Probably not in any measurable way. A 16-hour fast ends about where human glycogen stores are still being used up; the metabolic switch to ketones that precedes a meaningful autophagy signal usually completes after 24 hours or more. Time-restricted eating still improves insulin, appetite and weight, but no human study has shown it raises autophagy.
Is a 40-hour fast enough for autophagy?
Possibly. Human data at 24 to 72 hours of water fasting is limited to small studies of autophagy-related proteins in blood cells, which are suggestive rather than definitive. By 40 hours most people are in sustained ketosis with mTOR and IGF-1 suppressed, which is the state in which autophagy is expected to rise. The only direct human measurement of autophagic flux comes from a five-day fasting-mimicking diet.
How many hours of fasting does it take to trigger autophagy?
There is no validated hour count for humans. Mouse studies show autophagy rising within about 24 hours, but mice switch fuels far faster than people. In humans, the signal appears with multi-day protocols: suggestive changes at 48 to 72 hours, and a roughly doubled autophagic flux by day five to six of a fasting-mimicking diet.
What is the difference between a fasting-mimicking diet and water fasting for autophagy?
A fasting-mimicking diet is a five-day, plant-based, low-protein, low-sugar meal plan calibrated so the body reads it as fasting while you still eat, which reduces muscle loss, electrolyte problems and drop-out. Water fasting produces a deeper fasted state but needs medical supervision beyond 48 hours. The fasting-mimicking diet is the protocol with direct human autophagy data.
Who discovered autophagy, and why does fasting matter for it?
Yoshinori Ohsumi received the 2016 Nobel Prize in Physiology or Medicine for identifying the genes that control autophagy in yeast. Nutrient scarcity is the strongest natural trigger of the process, which is why fasting became the main non-drug way to try to raise it.
Who should not do multi-day fasts?
Anyone with a history of an eating disorder, anyone underweight, anyone pregnant or breastfeeding, and anyone with type 1 diabetes. People on insulin, sulfonylureas, SGLT2 inhibitors, GLP-1 medications or diuretics need medication adjustments first. Talk to a physician before starting, not after.
References
Bagherniya M, Butler AE, Barreto GE, Sahebkar A. “The effect of fasting or calorie restriction on autophagy induction: A review of the literature.” Ageing Research Reviews 2018;47:183–197. DOI: 10.1016/j.arr.2018.08.004
Espinoza SE, Park S, Connolly G, et al. “Effect of fasting-mimicking diet on markers of autophagy and metabolic health in human subjects.” GeroScience 2025. DOI: 10.1007/s11357-025-02035-4
Anton SD, Moehl K, Donahoo WT, et al. “Flipping the Metabolic Switch: Understanding and Applying the Health Benefits of Fasting.” Obesity 2018;26(2):254–268. DOI: 10.1002/oby.22065
de Cabo R, Mattson MP. “Effects of Intermittent Fasting on Health, Aging, and Disease.” New England Journal of Medicine 2019;381:2541–2551. DOI: 10.1056/NEJMra1905136
