Five-Day Fast: Who Keeps the Weight Off 12 Weeks Later?

A randomized trial published on 16 September 2026 in Genome Medicine found that a five-day fast lowered body mass index, mostly through loss of fat mass, and that the improvement only partially persisted 12 weeks later. The researchers then built a model from each person's starting gut bacteria and clinical values that predicted who held on to more of the change. That model is an early research result, not a test you can take.

Safety first. A five-day fast is an extended fast. The one studied here was carried out inside a clinical trial with healthy adults. If you are pregnant or breastfeeding, have diabetes or take blood-sugar medication, have a history of eating disorders, take other regular medication, or are under 18, talk to a doctor before fasting. Do not start a multi-day fast without medical advice, and stop if you feel unwell.

What the study did

The trial is called the LEANER study. It was registered in advance, on June 29, 2020.

Who	38 healthy adults who completed the intervention
Design	Randomized, waitlist-controlled trial
Intervention	A 5-day fast
Follow-up	12 weeks
Measured	Body mass index and body composition, gut microbiome composition, plasma and fecal metabolites

"Waitlist-controlled" means participants were randomly assigned either to fast first or to wait, so that the waiting group could serve as a comparison. That is a stronger design than simply measuring people before and after.

The second half of the study is the prediction work. The team used machine learning with cross-validation to ask whether data collected at baseline, before anyone fasted, could predict body weight response at 12 weeks. They then checked the model against three independent groups of people who had done prolonged fasts in other studies.

What it found

Weight and fat. Fasting reduced body mass index acutely, "predominantly driven by loss of fat mass". At 12 weeks, these improvements "partially persisted". The abstract gives no figures in kilograms or BMI points for either time point, so we give none. What you can take from the wording is the direction: some of the change lasted, some did not.

Gut bacteria and metabolites. The fast produced marked shifts in the composition of the gut microbiome and in metabolites in blood and stool. How much a person's microbial diversity changed, right after the fast and later, was associated with how diverse their microbiome was at the start.

The predictors. A model combining baseline microbiome and clinical variables predicted BMI response at 12 weeks. The abstract names four prominent predictors:

• an unclassified Faecalibacterium species
• Oscillibacter sp. 50_27
• low-density lipoprotein (LDL) cholesterol
• systolic blood pressure

The abstract does not say in which direction each predictor works, so it would be a guess to say that more or less of any of them is "good".

It held up elsewhere. The authors report that the model generalized to three independent cohorts: people with metabolic syndrome, patients with multiple sclerosis exposed to repeated fasting, and healthy volunteers fasting for 6 to 12 days. Those were research cohorts fasting within studies, not people fasting on their own.

Why the same fast gives different results

Anyone who has fasted alongside a friend has seen this. Two people do the same thing for the same number of days, and 12 weeks on, one is lighter and the other is back where they began.

The usual explanation is behavior: what each person ate afterwards. This study points to an additional possibility, that some of the difference is visible before the fast starts, in the bacteria a person carries and in ordinary clinical values such as LDL cholesterol and blood pressure.

That is an appealing idea, and it fits the authors' conclusion that baseline characteristics "can help stratify expected longer-term responses". But notice the careful verb. Stratify means sorting people into likelier and less likely responders. It does not mean knowing the outcome for an individual.

What it does not show

• It is a small trial. 38 people is enough to detect clear average changes. It is a thin base for a prediction model, because with many microbiome variables and few people, models can latch on to patterns that are partly chance. The external validation in three cohorts is the right safeguard and makes the result more credible, but it does not turn it into a clinical tool.
• A prediction model is a hypothesis. The model shows that the signal exists in these datasets. Nobody has yet used it to choose who should fast and then checked whether those people did better. Until that kind of study is done, it remains a hypothesis.
• You cannot take this test. A consumer stool test will not tell you whether a five-day fast will "work" for you. The predictors came from research-grade sequencing combined with clinical measurements inside a statistical model.
• Prediction is not cause. Even if those bacteria reliably mark people who keep weight off, the study does not show that changing them would change the outcome.
• Healthy adults only in the main trial. The LEANER participants were healthy. The results do not tell people with medical conditions what a five-day fast would do for them.
• 12 weeks is not long. "Sustained" here means 12 weeks of follow-up. What happens after that was not measured.

What this means if you fast

The honest takeaway is about expectations. In a controlled trial, the average result of a five-day fast was a change that partly faded over 12 weeks. If that has been your experience, it is the typical pattern, not a personal failure.

It also suggests that comparing your result with someone else's is not very informative. People start from different places.

What you do control is the unglamorous part: preparing properly, staying on top of fluids and electrolytes, and coming back to food carefully. Our guides on structuring longer fasts and on hydration and electrolytes during extended fasts cover those steps. If you are new to longer fasts, a shorter one such as a 72-hour fast already calls for the same care and the same conversation with a doctor.

QoFast is a fasting tracker app for iPhone and Android. It supports extended fasts of 3, 5 or 7 days or a custom length, each with an optional preparation phase and a refeeding phase, and it keeps a history of your fasts. It gives general information, not medical advice. It is free to download, with an optional Pro subscription: App Store and Google Play.

FAQ

How much weight do you lose on a five-day fast? The abstract of this trial does not give a figure. It reports that body mass index fell, mostly from fat mass, and that the change partially persisted at 12 weeks.

Does the weight come back after a five-day fast? On average, part of it did in this trial. The improvements "partially persisted" at 12 weeks, which means some of the initial change was lost.

Can a gut microbiome test tell me if fasting will work for me? Not on the basis of this study. The model was built in 38 people and checked in three research cohorts. It is a research finding that needs prospective testing, not a product.

Which gut bacteria predicted the result? The abstract names an unclassified Faecalibacterium species and Oscillibacter sp. 50_27 among the prominent predictors, alongside LDL cholesterol and systolic blood pressure.

Sources

• Kaufhold G, Bartolomaeus TUP, et al. "Machine learning identifies microbiome and clinical predictors of sustained weight loss following prolonged fasting." Genome Medicine, 16 September 2026. DOI: 10.1186/s13073-026-01765-0, PubMed 42754908