Intermittent fasting produces weight loss. So does eating the same number of calories across a normal day. The question that decides whether fasting is worth anything is which of those two produces more, and the randomized trials that answer it mostly find a small difference or none. That comparison is the entire story, and almost every enthusiastic account of fasting skips it by comparing fasting against no intervention at all.
The article that stood here made three claims worth retracting. Fasting does not “prime the metabolism” or “jump start” fat burning. Sirtuins are enzymes, not receptors, and no human trial has shown that fasting extends human life. And a 35-calorie serving of anything is not something you consume while fasting.
Why the comparison group decides everything
A fasting protocol is a rule about timing. It restricts when you eat, not what or how much, and any weight loss it produces has to arrive through the same route as every other diet — a smaller total intake.
So a trial that compares 16:8 eating against a control group doing nothing is measuring whether skipping breakfast makes people eat less. That is a real and useful question. It is not the question of whether fasting has an advantage over simply eating less, which requires matching calories or matching the level of support in both arms. Those trials exist, and they are less flattering.
Time-restricted eating: what the trials found
The sharpest test is TREAT, published in JAMA Internal Medicine in 2020. Lowe and colleagues randomized 116 adults with overweight or obesity to either eating freely between noon and 8pm, or three structured meals a day, for 12 weeks. The fasting group lost 0.94 kg and the structured-meals group 0.68 kg. The between-group difference was 0.26 kg, with a confidence interval running from −1.30 to +0.78 kg. Nothing separated them.
A longer trial in the New England Journal of Medicine in 2022 followed 139 adults in China for 12 months, with every participant on the same calorie target and one group additionally confined to an 8am–4pm window. The time-restricted group lost 8.0 kg and the calorie-restriction-only group 6.3 kg — a 1.8 kg difference that did not reach significance, with no separation in blood pressure, waist circumference or lipids.
A 2023 trial in Annals of Internal Medicine took a different angle, testing whether an eating window could substitute for counting. Ninety adults, 33 percent Black and 46 percent Hispanic, were assigned to noon-to-8pm eating with no counting, a 25 percent calorie reduction, or control, for 12 months. Time-restricted eating produced 4.61 kg more loss than control and calorie restriction 5.42 kg; the two active arms did not differ significantly.
A 2025 systematic review in the International Journal of Behavioral Nutrition and Physical Activity pooled 20 randomized trials and 1,242 participants. Time-restricted eating alone beat no restriction by 1.59 kg. Added on top of energy restriction, it contributed a further 0.94 kg — real, but modest, and the authors concluded that combining the two offered no meaningful additional benefit.
Alternate-day and 4:3 fasting
Harsher protocols have been tested against daily restriction directly. Trepanowski and colleagues, in JAMA Internal Medicine in 2017, randomized 100 adults to alternate-day fasting, daily calorie restriction or control for 12 months. At six months both active arms had lost 6.8 percent of body weight. At 12 months the fasting group was at −6.0 percent and the restriction group at −5.3 percent, with no significant difference and no separation in fat mass, lean mass, visceral fat or cardiovascular markers. Dropout was highest in the fasting arm at 38 percent, against 29 percent for daily restriction.
The most interesting recent result runs the other way. A 2025 trial in Annals of Internal Medicine randomized 165 adults to 4:3 fasting — an 80 percent energy restriction on three non-consecutive days a week — or daily calorie restriction, both with gym membership and behavioral support, for 12 months. The fasting group lost 7.6 percent of body weight against 5.0 percent, a 2.6-point difference favoring fasting, with better changes in blood pressure, cholesterol and fasting glucose.
The researchers’ own explanation is the sober one: fasting may win when it wins because three hard days a week is easier to keep track of than seven moderate ones. That is an adherence effect, which is exactly what the broader diet-trial literature keeps identifying as the deciding variable.
| Trial | Design | Fasting arm | Comparison arm | Difference |
|---|---|---|---|---|
| TREAT, 2020 | 116 adults, 12 weeks | −0.94 kg | −0.68 kg (3 meals) | Not significant |
| Liu et al., 2022 | 139 adults, 12 months | −8.0 kg | −6.3 kg (same calories) | Not significant |
| Lin et al., 2023 | 90 adults, 12 months | −4.61 kg vs control | −5.42 kg (25% cut) | Not significant |
| Trepanowski, 2017 | 100 adults, 12 months | −6.0% | −5.3% (daily cut) | Not significant |
| 4:3 trial, 2025 | 165 adults, 12 months | −7.6% | −5.0% (daily cut) | 2.6 points, favors fasting |
The lean-mass question
TREAT’s most cited result was not its weight outcome. In a 50-person in-person subgroup measured by DXA, the fasting group lost 1.70 kg, of which roughly 1.10 kg — about 65 percent — was fat-free mass. The usual expectation for weight loss is 20 to 30 percent. Appendicular lean mass index differed significantly between groups, favoring the structured-meals arm.
That subgroup was small and the finding has not been cleanly replicated, so it should not be treated as settled. But it is not isolated: the 2025 meta-analysis above found fat-free mass fell by about 0.58 kg with time-restricted eating and 0.56 kg when it was added to energy restriction. Compressing the eating window tends to compress protein intake with it.
The practical implication is unremarkable and applies to any weight-loss approach: adequate protein spread across the day, and resistance training, are what protect lean tissue. The protein dose-response evidence is covered separately.
Who should not fast
This is the part of the topic where getting it wrong causes real harm, so it is stated plainly rather than hedged.
- Anyone with a history of an eating disorder, or current disordered eating. A study of more than 2,700 Canadian adolescents and young adults published in Eating Behaviors found intermittent fasting reported by 47 percent of women and 38 percent of men in the past year, and among women it was associated with every measured disordered-eating behavior, including binge eating, vomiting and compulsive exercise. Association is not causation, but a rule that legitimizes long periods of not eating is a poor fit for anyone whose relationship with food is already strained.
- People with type 1 diabetes, and anyone taking insulin or a sulfonylurea. These medications carry hypoglycemia risk that fasting periods change substantially. Doses generally need adjusting in advance under clinical supervision, and the trial evidence in type 1 diabetes amounts to little more than pilot data.
- Anyone pregnant or breastfeeding. There is no adequate safety evidence, and energy and micronutrient requirements are elevated.
- Children and adolescents, and anyone underweight or with a history of being underweight.
- Anyone on medications that must be taken with food, or with a condition requiring regular eating.
If any of that applies, the right next step is a conversation with a physician or registered dietitian, not a modified protocol found online. For concerns about disordered eating specifically, the National Eating Disorders Association maintains screening tools and treatment referrals, and eating disorders are treatable conditions with genuine clinical pathways.
Where the evidence is weak
Most fasting trials are small. TREAT enrolled 116 people and reported its body-composition result from 50. Trepanowski enrolled 100. These are not sample sizes that settle questions, and the 2025 4:3 result, at 165 participants, is a single trial awaiting replication.
Most trials are also short. Twelve weeks is common; 12 months is rare; nothing runs for a decade. Adherence is self-reported in nearly every case, meaning the actual eating window is unverified. Dropout in the fasting arms tends to run higher than in the comparison arms, which biases per-protocol results toward whoever tolerated the protocol.
The longevity claims deserve their own caution. The best human data is the CALERIE trial, which randomized about 220 healthy adults to two years of calorie restriction and reported a 2 to 3 percent slowing on an epigenetic measure of the pace of aging (Nature Aging, 2023). Lifespan was not measured, because measuring human lifespan is not practical in a trial. Extrapolating from mouse studies and biomarker shifts to human longevity is not supported.
If you want to try it
For people with no contraindication, time-restricted eating is a reasonable option — not because it does anything special, but because a simple rule is easier to follow than an arithmetic one for some people, and simplicity is a legitimate advantage.
A 10- or 12-hour window is a gentler starting point than 8. The total amount eaten still governs the result, for the reasons set out in the piece on energy balance and calorie counting. Whatever the window, hitting a protein target inside it takes deliberate effort. Hydration should continue through the fasting hours. Caffeine on an empty stomach is worth watching if it makes you jittery or disrupts sleep.
Two signs it is not working: eating in the window becomes compensatory rather than normal, or thinking about the next meal starts crowding out other things. Either is a reason to stop, and neither is a failure of discipline.
Common questions
Is 16:8 better than just eating fewer calories?
The randomized evidence says not by much. TREAT found a 0.26 kg difference over 12 weeks against structured meals, the 2022 NEJM trial found 1.8 kg over a year with no statistical separation, and the 2023 Annals trial found no significant difference against a 25 percent calorie cut. The advantage, where one exists, appears to be about ease rather than physiology.
Does fasting put you in “starvation mode”?
No. Energy expenditure does fall as weight falls, but that happens with any method of losing weight and is a matter of a few hundred calories a day rather than a metabolic shutdown. TREAT found no significant difference in resting metabolic rate between its arms.
Will I lose muscle if I fast?
Possibly more than with other approaches, though the evidence is thin. The TREAT subgroup found about 65 percent of lost weight came from fat-free mass, and pooled trials show fat-free mass falling by roughly half a kilogram. Adequate protein and resistance training are the countermeasures, and they matter for any weight-loss method.
Can I fast if I take medication for type 2 diabetes?
Not without talking to whoever prescribes it first. Insulin and sulfonylureas in particular carry hypoglycemia risk that fasting changes, and doses usually need adjusting before rather than after starting. Other medications are less affected, but that is a judgment for a clinician who knows your history.
Where this leaves it
Intermittent fasting is a scheduling tool with a modest evidence base and a clear list of people it does not suit. Against nothing, it works. Against the same calories eaten differently, it mostly draws. That is a defensible reason to try it and a poor reason to be evangelical about it.
Greens Plus makes green superfood powders and bars; they contain calories and are food, not something to consume during a fasting window. What is in them is listed at greensplus.com.
About the Author
