Fasting Is Worth Doing. The Reasons Everyone Gives Are Mostly Wrong

Cochrane found almost nothing, and said so out loud

In February 2026 the Cochrane Collaboration published the most skeptical assessment intermittent fasting has ever gotten from a serious source. Twenty two randomized trials, 1,995 adults. Against ordinary dietary advice, fasting produced a difference of 0.33% of body weight. On a 200 pound person that's about ten ounces. The confidence interval crossed zero and the certainty was rated low.

The rest is no kinder. The odds of reaching a clinically useful 5% weight loss came out identical. Against no intervention at all, fasting won by 3.42% of body weight, a result the reviewers themselves called "unlikely to be clearly clinically meaningful." Quality of life was measured in three trials and improved in none.

Lead author Luis Garegnani put it about as bluntly as review authors ever do: "Intermittent fasting just doesn't seem to work for overweight or obese adults." Online enthusiasm, he added, "may be running ahead of the data."

He's right about the enthusiasm. We still come out in favor of fasting. Those aren't in conflict, and the reason is the rest of this piece.

The strongest trial in fasting's favor is really about sticking with it

DRIFT, published in the Annals of Internal Medicine in 2025 out of the University of Colorado Anschutz, ran 165 adults for a year. One group did 4:3 fasting: 80% energy restriction on three nonconsecutive days a week. The other cut calories 34.3% every day. At twelve months the fasting group had lost 7.7 kg against 4.8 kg, with higher adherence at six and twelve months.

Notice what that is and isn't. It isn't metabolic magic. Both groups ate less. The fasting group just did more of what it was asked to do, for longer.

An 18-month Adelaide trial in Clinical Nutrition this year, under Prof. Leonie Heilbronn, makes the same point in the participants' own words. Over 200 adults with obesity; fasting and calorie restriction both took off about 7 kg at six months against 2 kg for standard care. The fasting group didn't feel they had to constantly monitor their eating or count calories.

The BMJ ran a network meta-analysis in June 2025 covering 99 trials and 6,582 adults. Alternate day fasting came out 3.40 kg ahead of unrestricted eating at high certainty, and beat continuous calorie restriction by 1.29 kg at moderate certainty. The part that never makes the carousel: in the 17 trials that ran 24 weeks or longer, the advantage vanished.

So fasting's edge is real, modest, and fragile. It shows up while people can follow the rules and it fades when they stop.

The mechanism is boring, which is why we believe it

Gabel and Varady at the University of Illinois Chicago documented what actually happens when you compress eating into a 6 to 8 hour window: people spontaneously eat 350 to 500 fewer calories a day. Nobody tells them to. Nobody counts anything.

That's the trick. All of it.

It's an unglamorous explanation and it's far more persuasive than the autophagy story. It explains why fasting works, and it explains why fasting doesn't beat careful calorie counting, because it is calorie counting, outsourced to a clock instead of a food scale.

The hype runs in both directions, and both directions are thin

Yoshinori Ohsumi won the 2016 Nobel Prize for autophagy. He won it for work done in yeast. Everything since, the day counts, the hour marks, the graphics announcing that your cells start eating themselves at hour 16, is extrapolation.

Researchers have measured autophagy markers like ATG5 and LC3-II in people fasting for Ramadan, and the markers do move. What nobody has done is connect fasting-induced autophagy in a human being to lifespan, to disease, or to any outcome you'd care about. One study found intermittent fasting switched on autophagy markers in mouse liver but not in mouse or human muscle. And no study has established a day, or an hour, at which human autophagy peaks. When someone tells you day four is when it happens, they're quoting a podcast.

Now run the hype the other way. In March 2024 a headline traveled everywhere: eating windows under eight hours were linked to a 91% higher risk of cardiovascular death across roughly 20,000 American adults. It was an American Heart Association conference abstract, never peer reviewed, observational, built from as little as two 24-hour dietary recalls spread over a decade. The obvious problem sits right in the design: people with cancer and other advanced illness lose their appetite and eat in shorter windows. That abstract isn't strong enough to frighten you, for the same reason the autophagy graphics aren't strong enough to sell you a supplement.

Muscle loss is the downside with the best numbers behind it

The TREAT trial, in JAMA Internal Medicine in 2020, put 116 adults on 16:8 for 12 weeks. The weight loss was trivial and not significant. What made the paper matter was the body composition: of the 1.70 kg the fasting group lost, 1.10 kg was lean mass. About 65%, against a normal share of 20 to 30%. The authors wrote that this "serves as a caution for patient populations at risk for sarcopenia."

That should change how you fast, not whether you do. The fix is well supported: resistance training plus enough protein. Systematic reviews found people eating only 1.0 g of protein per kg of body weight a day lost lean mass even while lifting, while higher intakes held onto it. If you fast and the heaviest thing you lift all week is the groceries, you're choosing to lose muscle.

The medication section is the one to take literally

Here's where "talk to your doctor" stops being a formality.

The American Diabetes Association's 2025 Standards of Care name hypoglycemia as the major risk for anyone taking insulin, sulfonylureas, or meglitinides. Basal insulin, the ADA is explicit about this, should not be stopped even when you aren't eating. Published dose-adjustment guidance runs to cutting sulfonylureas by 50 to 100%, long-acting insulin by up to 20%, and short-acting by 10 to 50%. We're printing that so you know a protocol exists, not so you can apply it yourself. It belongs to whoever wrote your prescription.

SGLT2 inhibitors carry a different problem: euglycemic diabetic ketoacidosis, which can develop during a fast while your blood sugar readings look normal. That's the combination that catches careful people out.

Fasting and disordered eating share too much furniture

A 2022 study in Eating Behaviors surveyed about 2,700 Canadians aged 16 to 30. In the past year, 47% of women, 38% of men, and 52% of transgender and gender non-conforming respondents had fasted. Among the women, fasting was associated with every disordered eating behavior the researchers measured: binge eating, vomiting, laxative use, compulsive exercise. Among the men, with compulsive exercise.

It's observational, so it can't tell you which came first. Fasting may be the symptom rather than the cause. But a practice that hands you socially approved cover for not eating isn't neutral for everyone.

Who should skip this entirely

The contraindication list in the literature is short and specific. People who are underweight or malnourished. Pregnant women. Anyone with a history of an eating disorder. People with chronic kidney disease. Older adults at risk of frailty or falls. Anyone on warfarin or lithium. Add anyone taking the diabetes medications above who hasn't talked to their prescriber yet.

Gallstones belong on the list too: windows of 16 hours or more slow gallbladder emptying, and bile that sits still is how stones form.

The reassuring safety data comes from buildings with lab equipment

Longer fasts have a better safety record than most people expect. A 2019 PLOS ONE paper covering 1,422 people on Buchinger fasts of 4 to 21 days recorded adverse effects in under 1% and an absence of hunger in 93.2% of participants. A 2024 Nutrients trial at the TrueNorth clinic logged none that were serious or life-threatening.

Read that list again and notice what every entry has in common. Medically supervised. Residential. Daily clinical monitoring and blood work. That is not what's happening at your kitchen table, and it's the most important caveat in the subject. If you're going to do it anyway, and plenty will, the preparation and the refeed are where the risk lives, which is why we handled those separately in our 48-hour fast guide and the longer five-day version.