Every other day, and the next day only partly makes up for it#
Alternate-day fasting is the largest dose of intermittency anyone has tested at scale: eat nothing (or close to it) on one day, eat freely on the next, repeat. It works, and the reason it works is a single number. If people fully compensated on feed days, the schedule would remove nothing; if they didn't compensate at all, it would remove half their intake. In the best-measured version — 36 hours of zero calories alternating with a 12-hour eating window, in 60 healthy non-obese adults over four weeks — strict alternate-day fasting produced an average calorie reduction of 37%1.
That figure is doing a lot of work. Thirty-seven percent against a theoretical fifty means feed days ran at roughly 126% of an ordinary day's intake — people given free rein after 36 hours without food ate about a quarter more than usual, not double. (The 126% is my arithmetic on the authors' 37%, not a number they report.) So the deficit is real and large, and it survives contact with appetite. What it does not survive, on the current evidence, is time — and the protocol has a second problem the pooled statistics can't see. This is the most demanding schedule under the intermittent fasting umbrella, and the case for and against it is unusually concrete.
Alternate-day fasting doesn't work because a fast day is heroic. It works because the day after a fast day is only about a quarter bigger than an ordinary one.
The zero-calorie original, and the hunger that never left#
The founding human trial is small and worth reading in full detail, because its conclusion is not the one people quote.
Sixteen non-obese adults — eight men, eight women — fasted every other day for 22 days, with body composition, resting metabolic rate, respiratory quotient and blood markers measured at baseline, day 21 and day 22. They lost 2.5 ± 0.5% of body weight (P < 0.001) and 4 ± 1% of fat mass (P < 0.001). Fasting insulin fell 57 ± 4% (P < 0.001). Resting metabolic rate did not change significantly, and respiratory quotient fell on day 22, implying an average daily increase in fat oxidation of at least 15 grams2.
Then the finding that governs everything else: hunger rose on the first fasting day and stayed elevated (P < 0.001) for the entire three weeks. The authors' own conclusion was that this indicates "the unlikelihood of continuing this diet for extended periods of time," and they proposed a fix in the same sentence — adding one small meal on fasting days to make the approach more acceptable.
That sits awkwardly beside the alternate-day trials that measured appetite repeatedly and found it flat or falling, which are collected in fasting and hunger adaptation. It is not a contradiction so much as a dose difference, and the field's next move made that explicit.
Add one small meal, and it becomes a different protocol#
What almost everyone means by ADF today is the modified version: not zero, but roughly a quarter of your energy needs on fast days — about 500 calories, usually as one meal.
In a 12-week randomized trial, non-obese adults ate 25% of energy needs on fast days and ad libitum on feed days. Thirty completed, of 32 enrolled. The fasting group lost 5.2 ± 0.9 kg — 6.5% of body weight (P < 0.001) — with fat mass down 3.6 ± 0.7 kg and fat-free mass unchanged. Triacylglycerols fell 20%, C-reactive protein 13%, leptin 40%; adiponectin rose 6%. Self-reported adherence on fast days was 98 ± 5%, and hunger "remained moderate" while satisfaction and fullness increased3.
The hunger result is the opposite of Heilbronn's, and the separator is exactly the variable Heilbronn nominated: one small meal. A 500-calorie fast day is a different physiological and psychological event from a zero-calorie one, and the two literatures should not be pooled casually. Two cautions belong with the modified-ADF numbers, though. Thirty completers is small, this came from the laboratory that has championed the protocol for over a decade, and a self-reported 98% adherence rate is the kind of figure that improves when participants are recording their own compliance. That the lean mass held is genuinely encouraging, but it rests on one 12-week trial in thirty people.
Ranked first at three months, and gone by twelve#
Here is the live disagreement about ADF, and it resolves more cleanly than most.
A network meta-analysis of 47 randomized trials covering 3,363 adults ranked four calorie-restriction regimens against control. Alternate-day fasting produced the largest weight loss of any of them — −3.42 kg (95% CI −4.28 to −2.55) — ahead of time-restricted eating (−2.25 kg), short-term fasting (−1.87 kg) and continuous daily restriction (−1.59 kg)4. Sixteen of the included studies reported adverse events; none were serious, and ADF's side effects were described as fewer, shorter and less frequent than the alternatives'.
Against that sits the year-long randomized trial that found alternate-day fasting no better than plain daily restriction, with the highest dropout of any arm — the full accounting is in does intermittent fasting actually beat regular dieting. A pooled ranking that puts ADF first and a single well-run one-year trial that puts it level look like a straight contradiction.
They aren't, and the reviewers hand you the reconciliation themselves. Their own subgroup analysis found that all four regimens produce modest loss at one to three months with ADF ranked highest, that varying degrees of weight regain set in by four to six months, and that time-restricted eating potentially ranks first over both the 4–6 and 7–12 month windows. The advantage is real and it is a short-horizon advantage. Twelve-month trials are precisely where it stops being visible.
| Horizon | What ADF looks like |
|---|---|
| 1–3 months | Largest weight loss of any restriction regimen4 |
| 4–6 months | Regain sets in across regimens; TRE ranks ahead4 |
| 12 months | No advantage over daily restriction; highest dropout5 |
The reviewers also flag their own limitations plainly: inadequate blinding, small samples in many trials, and high dropout in some. A ranking built partly on people who finished is a ranking of a protocol among those who could tolerate it.
What that leaves ADF actually good for#
Read the whole picture and a specific, narrow use emerges rather than a verdict.
ADF is the most efficient way to open a large deficit quickly, and the fast-day mechanism is not willpower over food quantity — it is the removal of eating decisions for a whole day at a time. For someone who finds moderating every meal exhausting and would rather run a binary rule, that is a genuine fit, and the weekly-total logic that makes it legitimate is the same one behind 5:2, covered in does your body count calories daily or weekly.
What it is not is a protocol to plan a year around. The horizon on which it out-performs is roughly the first three months, its dropout rate at twelve is the worst on record, and the founding trial found hunger that never adapted at zero calories. If you use it, use the modified version — that 500-calorie fast-day meal is not a softening of the protocol, it is the version with the better evidence behind it — and treat it as a starting sprint you intend to hand off to something gentler, rather than the shape of your eating from here on. Keep protein high on fast days, expect the first fast day to be the worst one, and check who should not fast first: on a schedule this aggressive, the medication and training-load interactions stop being theoretical.
FAQ#
Do you eat nothing at all on an alternate-day fasting fast day?#
Both versions exist and they behave differently. The original trials used true zero-calorie fast days and found hunger that rose on day one and stayed elevated for three weeks2. Modern modified ADF allows about 25% of energy needs — roughly 500 calories in one meal — and in that form hunger stayed moderate while fullness rose3. The modified version is what most research now means by ADF.
Do you eat twice as much the day after a fast day?#
No, and that's why the protocol works at all. Four weeks of strict alternate-day fasting produced a 37% average calorie reduction rather than the 50% perfect compliance would give1 — implying feed days ran about a quarter above a normal day, not double it. Compensation is real but partial, and the leftover gap is the entire deficit.
How long does alternate-day fasting keep working?#
Its measurable edge is a short one. Pooled across 47 trials, ADF produced the largest weight loss of any restriction regimen at one to three months, but regain set in by four to six months and it no longer ranked first beyond that4. At twelve months it matched ordinary daily dieting while losing more participants along the way5.
Sources#
- Stekovic S, Hofer SJ, Tripolt N, et al. Alternate Day Fasting Improves Physiological and Molecular Markers of Aging in Healthy, Non-obese Humans. Cell Metab. 2019;30(3):462-476.e6.
- Heilbronn LK, Smith SR, Martin CK, Anton SD, Ravussin E. Alternate-day fasting in nonobese subjects: effects on body weight, body composition, and energy metabolism. Am J Clin Nutr. 2005;81(1):69-73.
- Varady KA, Bhutani S, Klempel MC, et al. Alternate day fasting for weight loss in normal weight and overweight subjects: a randomized controlled trial. Nutr J. 2013;12:146.
- Huang J, Li Y, Chen M, Cai Z, Cai Z, Jiang Z. Comparing caloric restriction regimens for effective weight management in adults: a systematic review and network meta-analysis. Int J Behav Nutr Phys Act. 2024;21(1):108.
- Trepanowski JF, Kroeger CM, Barnosky A, et al. Effect of Alternate-Day Fasting on Weight Loss, Weight Maintenance, and Cardioprotection Among Metabolically Healthy Obese Adults: A Randomized Clinical Trial. JAMA Intern Med. 2017;177(7):930-938.
