Intermittent fasting produces weight loss, but trials so far show it works about as well as simply cutting calories: a 2022 year-long randomized trial in the New England Journal of Medicine found that time-restricted eating with calorie restriction produced roughly 8 kilograms of weight loss versus about 6.5 kilograms with calorie restriction alone — a difference that did not reach statistical significance. Whether fasting schedules offer anything beyond conventional dieting, and whether they are safe over decades, remain open research questions as of early 2026.
Newspaper Daily publishes information, not medical advice. Fasting is not appropriate for everyone, and people who take glucose-lowering medication, are pregnant, have a history of eating disorders, or manage chronic illness should talk with a physician before changing meal timing.
What are the main forms of intermittent fasting?
Research literature describes several distinct protocols that are often lumped together in popular coverage:
- Time-restricted eating: all calories within a window of roughly 4 to 10 hours daily, with water-only fasting the rest of the day; the 16:8 schedule is the best-known version
- Alternate-day fasting: alternating fast days, sometimes allowing about 25 percent of energy needs, with normal eating days
- The 5:2 pattern: five normal eating days and two nonconsecutive low-calorie days of about 500 to 600 calories weekly
- Periodic prolonged fasting: fasting for two or more consecutive days, studied mainly in small trials and animal work
These protocols differ in adherence difficulty, physiology, and evidence quality, which is why conclusions drawn about one cannot simply be transferred to the others.
How much weight do people lose in trials?
Meta-analyses of randomized trials converge on modest amounts. A 2020 meta-analysis in JAMA Network Open of 27 fasting trials found average weight loss of roughly 0.8 to 1.5 kilograms per week initially, settling around 4 to 8 percent of body weight over longer follow-up. Direct comparisons against continuous calorie restriction repeatedly find no meaningful difference: the 2022 Chinese year-long NEJM trial of time-restricted eating, the 2020 TREAT trial of alternate-day fasting, and a 2022 NEJM comparison of alternate-day versus daily restriction all reported statistically indistinguishable weight loss between fasting and conventional dieting arms.
Adherence, not fasting chemistry, appears to drive results. In TREAT, the alternate-day arm ate more on feast days, effectively neutralizing the intended deficit — an illustration that a fasting schedule only works if the net calorie reduction actually happens.
Does fasting improve metabolism beyond weight loss?
Trials report mixed metabolic findings. The 2022 NEJM year-long trial found no significant additional benefits in insulin sensitivity, blood pressure, or lipid levels beyond what weight loss itself explained. A 2018 Cell Metabolism trial of early time-restricted eating — eating only between roughly 8 a.m. and 2 p.m. — found improved insulin sensitivity and blood pressure even without weight change, an intriguing result from a small, short study. Work in animals showing longevity and cellular-repair benefits from fasting has not translated cleanly to humans; as researchers reviewing the field in Nature Reviews Endocrinology have noted, most human trials are short, and calorie matching between arms often erases apparent metabolic advantages.
What about the heart-risk headlines?
In 2024, an abstract presented at an American Heart Association conference drew wide coverage after an analysis of NHANES data followed about 20,000 adults and reported that an 8-hour eating window was associated with a 91 percent higher risk of cardiovascular mortality over roughly 8 years. The finding was observational, based on two days of dietary recall, was not peer-reviewed at the time, and its authors themselves emphasized that meal timing was recorded only at baseline. Cardiologists quoted by Reuters and statnews.com in March 2024 urged caution in interpreting it, and the abstract's exposure measurement makes confounding by illness and eating-pattern instability plausible. The episode is best read as a reason for longer-term trial data, not as established risk.
Related stories: Mediterranean Diet: What the Evidence Actually Shows · Omega-3 Supplements and Heart Health: What Large Trials Show.
Who was fasting designed for, and who should not do it?
Insulin-requiring and sulfonylurea-treated diabetes is the clearest exclusion without medical supervision: fasting schedules in these patients have caused hypoglycemia, and clinical endocrinology guidance treats medication adjustment as a prerequisite. Pregnancy and breastfeeding, adolescence, a history of anorexia or bulimia, and underweight are additional populations where professional organizations advise against unsupervised fasting. A 2023 Annals of Internal Medicine randomized trial of time-restricted eating without calorie restriction did find modest weight loss — about 3.3 kilograms at six months — but the same journal's commentary stressed that individual medical screening preceded enrollment.
What happens to muscle during fasting?
Concerns are legitimate but manageable in trial settings. Reviews of fasting trials report lean mass typically accounts for one-quarter to one-third of total weight lost — a proportion similar to ordinary dieting. Trials that paired fasting with adequate protein and resistance exercise found better lean mass retention, and a 2021 study of time-restricted eating in men doing resistance training found no loss of muscle strength. For older adults, however, where sarcopenia already threatens function, protein-sparing concerns make unsupervised fasting less attractive to geriatric clinicians.
Does the timing of the eating window matter?
Possibly. Circadian-biology research suggests earlier windows — finishing eating by mid-to-late afternoon — produce better glucose outcomes than identical windows shifted late into the evening. The 2018 early-window trial found this, and a 2022 randomized crossover study found late eating increased hunger and reduced energy expenditure relative to early eating. Human chrononutrition remains a young field, but the direction of small-trial findings has been consistent.
What is it actually like to follow?
Attrition and adherence data from trials suggest moderate difficulty: dropout in time-restricted eating trials has generally been comparable to other diets, while alternate-day fasting shows higher attrition in some studies. Commonly reported early effects include morning hunger, irritability, and headache in the first one to two weeks, per trial adverse-event tables; headaches, dizziness, and constipation were the most frequently reported symptoms in a 2022 systematic review. Social eating patterns — family dinners, breakfast culture — are the adherence obstacles clinicians report most often in practice.
When to see a doctor or registered dietitian
Medical input is essential before fasting for anyone on insulin or sulfonylureas, anyone who is pregnant or breastfeeding, adolescents and older adults, people with a personal or family history of eating disorders, and people with significant heart, kidney, or liver disease. During fasting, symptoms such as fainting, confusion, chest discomfort, or blood glucose readings below individualized targets warrant stopping and contacting a clinician. A registered dietitian can also help structure nutrient-adequate meals inside a shortened window, since compressed eating periods make fiber, protein, and micronutrient targets harder to hit.
The bottom line on intermittent fasting evidence
The randomized evidence supports a straightforward summary: intermittent fasting works about as well as calorie restriction for weight loss, with no clear metabolic bonus in the best-controlled year-long trial; benefits appear largest when the eating window lands earlier in the day; and the 2024 observational cardiovascular-mortality finding is a caution signal, not a verdict. Long-duration trials — years, not months — remain the missing piece as of March 2026.