Skip to content
HabitSeek
Back to results

Intermittent fasting: daily eating windows, alternate-day and 5:2 approaches

Restricting when or on which days you eat rather than how much: a daily eating window such as 16:8, whole days of severe energy restriction such as alternate-day fasting or two low-calorie days a week, or a combination. These patterns are grouped here because people search for them together, not because the evidence shows they work the same way.

At a glance

  • Evidence: Every fasting pattern reduces weight against unrestricted eating, but the estimable effect is modest, variable by pattern, and mostly attributable to eating less overall.Body weight versus unrestricted eating, by fasting pattern · Limited certainty · Randomized intervention
  • What’s uncertain: Most trials are short, and fasting's advantages over ordinary calorie restriction shrink or disappear with longer follow-up.
  • First step: Choose a pattern you can maintain, and plan maintenance as well as fasting; year-long trials found adherence was a key constraint.
  • Before you try it: None of these patterns is recommended during pregnancy or breastfeeding, or for children, because safety has not been adequately studied in these populations. All precautions
Read the full takeaway

Weight outcomes

Eating windows, alternate-day fasting and 5:2 all work for weight loss, and the most useful finding about them is that none works better than simply eating less. This entry covers all three because people search for them together, not because the evidence shows they are equivalent: the trials compare them against different things, and the differences between the patterns are real but small. A 2025 network meta-analysis of 99 randomised trials with 6,582 adults found that every fasting strategy and continuous energy restriction reduced weight against an unrestricted diet, with alternate-day fasting the only fasting pattern that beat continuous energy restriction - by 1.29 kg (95% CI -1.99 to -0.59) - and the same paper labels that difference trivial, since it falls below the 2 kg threshold the authors treat as clinically meaningful. In trials lasting 24 weeks or more, no fasting strategy differed from continuous energy restriction at all.

Time-restricted eating and energy balance

Time-restricted eating is the same story with a smaller evidence base of its own: the largest pooling of it against unrestricted eating found about 1.4 kg of weight loss, and the two pivotal calorie-matched trials found no significant between-group difference at 12 weeks or 12 months. The mechanism is not fasting: 73 of the 99 comparisons involved a negative energy balance, and the review states that fasting's proposed metabolic advantages have not been substantiated in humans.

Risk factors and body composition

For cardiometabolic risk factors the picture is null almost everywhere, with HbA1c showing no difference in any of the ten comparisons and no consistent advantage over matched energy restriction, although eating-window timing does show timing-specific signals that have not been demonstrated for fasting days. Several syntheses report reduced fat-free and lean mass, which is a body-composition signal from secondary trial endpoints rather than a demonstrated harm, and the largest dedicated fat-free-mass meta-analysis did not reach significance for its own primary estimate.

Clinical outcomes

There is no randomised evidence at all on hard outcomes: no trial of any of these patterns has measured cardiovascular events, diabetes incidence or death, and the only hard-outcome data are observational and concern eating windows, where a U-shaped association with mortality was found and shorter windows were associated with higher mortality. Weight loss is therefore the outcome the evidence supports, and it should not be read as evidence of longer life, of clinical autophagy, or of a metabolic advantage over ordinary calorie restriction.

A fresh meal and glass of water with a modern kitchen clock nearbyDiscussed by5 peopleView statements ↓
NutritionLong-term healthVaries · Moderate · Free
Reviewed Oct 3, 2026
Save

Scientific evidence

What the research shows

Outcome assessments based on research, including its uncertainties and limitations.

Cardiovascular events, diabetes incidence and mortality, and the autophagy claim

Finding: Insufficient direct evidenceCertainty: Not assessedthis has not been directly tested

Evidence basis: indirect evidence, not a direct test of this exact habit

No randomised trial of any fasting pattern has measured cardiovascular events, diabetes incidence or death, so no longevity claim is available.

Population: No randomised trial of any fasting pattern has measured a clinical outcome. The observational data come from 33,052 US adults in a nationally representative cohort with a median eight years of follow-up, and they concern eating windows rather than fasting days. The autophagy evidence comes from mechanistic studies in humans and mice, including a trial involving 50 women with obesity and muscle biopsies.

Research details

Summary

No randomised evidence exists on hard clinical outcomes for any of these patterns. None of the 99 trials in the network meta-analysis measured a clinical event, and the 2026 Cochrane review found that none of its included studies reported diabetes status or an overall measure of comorbidity. The only hard-outcome data retrieved in this area are observational and concern eating windows rather than fasting days: a prospective cohort of 33,052 US adults followed for a median of eight years, with 4,158 deaths, found a U-shaped association between eating window and mortality with the lowest risk at about 11 to 12 hours a day, and higher all-cause mortality in people with windows of eight hours or less and of 15 hours or more. That is time-restricted eating, it is a cohort rather than a trial, and it cannot be transferred to alternate-day fasting or 5:2 - a point that matters more now that one entry covers both. On weight loss alone no claim about longer life is available: the trials that produced the weight loss lasted weeks to months and measured kilograms. The claim most often attached to fasting, that it promotes autophagy and thereby prevents disease, is not established in humans: direct measurement of autophagic flux found it unchanged between fasting and postprandial states in human blood cells, a 24-hour fast reduced autophagy markers in human muscle, and no human study has measured autophagy at 16 hours or linked it to any clinical outcome. The trials themselves do not support a clinical benefit: the longest alternate-day fasting trial found no cardioprotection, and the pooled weight difference at six months or more is small.

Why this certainty

Rated Not assessed because the outcome has never been measured in a randomised trial, so there is nothing to grade up or down. The basis is Indirect evidence because any claim about mortality or disease incidence has to be assembled from weight change, surrogate markers and observational data about one pattern among several. The outcome exists so that a reader who sees a benefit for weight also sees that the endpoints that would justify a longevity claim were never measured, and so that the autophagy claim is answered from human mechanistic work rather than from animal studies.

Limitations

This is an absence of evidence, not a negative result, and it is the most important limitation on how any of these patterns can be described: nothing in the randomised literature can tell a reader whether an eating window, alternate-day fasting or 5:2 changes cardiovascular events, diabetes incidence or death. The observational mortality data are for one pattern only and cannot be used to reassure or to warn about the others. The mechanism claims are weaker than they sound: the human autophagy evidence is a small mechanistic study measuring a marker in blood cells, a 24-hour fast reduced rather than increased markers in muscle, and animal work does not transfer. Because the weight-loss effect is small and mostly attributable to the energy deficit, extrapolating from it to mortality would require an assumption the evidence does not support.

Sources for this outcome:

Body weight versus unrestricted eating, by fasting pattern

Finding: Mixed evidenceCertainty: Limitedstudies disagree, or the effect depends on who is studied

Evidence basis: randomised trials

Every fasting pattern reduces weight against unrestricted eating, but the estimable effect is modest, variable by pattern, and mostly attributable to eating less overall.

Population: 6,582 adults across 99 randomised clinical trials in the network meta-analysis, of whom 720 were healthy and 5,862 had existing health conditions, with a median age of 45 years, median BMI 31 and median follow-up of 12 weeks; only 17 trials ran 24 weeks or more and only five reached 52 weeks. Twelve trials enrolled people with diabetes and 54 enrolled people with overweight or obesity. The time-restricted eating pooling covered 29 trials and 2,485 individuals. Trials were predominantly in high-income settings with adults who had overweight, obesity, metabolic syndrome or type 2 diabetes.

Research details

Summary

Against an unrestricted or usual diet, all of these patterns produce weight loss, and the size depends on the pattern and the pooling. The anchor is a 2025 network meta-analysis of 99 randomised clinical trials with 6,582 adults, median age 45, median BMI 31 and median follow-up 12 weeks, in which every fasting strategy and continuous energy restriction reduced body weight compared with an ad-libitum diet: alternate-day fasting by 3.40 kg (95% CI -4.14 to -2.67, high certainty), whole-day fasting by 2.36 kg, continuous energy restriction by 2.11 kg and time-restricted eating by 1.72 kg. Time-restricted eating has its own pooling against control: a meta-analysis of 29 randomised trials in 2,485 individuals found -1.37 kg (95% CI -1.99 to -0.75), from trials of which the authors judged 22 to have high risk of bias and none to be at low overall risk. On alternate days the advantage over a daily window is real but small: alternate-day fasting beat time-restricted eating by 1.69 kg (95% CI -2.49 to -0.88) and whole-day fasting by 1.05 kg (95% CI -1.90 to -0.19) in the network meta-analysis. The reason this outcome is recorded as Mixed rather than as a straightforward benefit is that the estimate depends entirely on the comparator and the pattern: against unrestricted eating the effect is consistent and positive, the pattern that produces the largest effect is the one with the worst adherence, and the mechanism the review itself identifies is the energy deficit rather than the fasting, with 73 of the 99 comparisons involving a negative energy balance and the authors stating that fasting's proposed metabolic advantages have not been substantiated in humans.

Why this certainty

Rated Limited and recorded as Mixed evidence. It is not Insufficient direct evidence, because randomised trials did measure these patterns against unrestricted eating and the direction is consistent across 99 trials and independent syntheses. It is not Moderate because the estimate is comparator-dependent and pattern-dependent, the largest effect comes from the pattern with the worst adherence, median follow-up is 12 weeks, the most favourable pooling contains no trial at low overall risk of bias, and the review's own authors attribute the effect to the energy deficit rather than to fasting. The label is Mixed rather than Benefit supported because a reader who takes 'intermittent fasting causes weight loss' from this outcome and then compares it with eating less will find the advantage gone, and the entry should not have to say that in a footnote.

Limitations

The trials are short: median follow-up is 12 weeks, only 17 of 99 ran at least 24 weeks, and the apparent advantages shrink or disappear in the longer subset. Most trials are small, blinding is impossible for a diet, attrition is high and differential, and analyses of completers flatter the more demanding patterns: dropout was highest in the alternate-day fasting arm of the twelve-month TREAT trial at 38% against 29% for daily restriction, and adherence to whole-day fasting in the network meta-analysis fell from about 74% at six weeks to about 22% at 52 weeks. The most favourable time-restricted-eating pooling judged no included trial to be at low overall risk of bias, so its estimate is inflated by design quality rather than only by chance. Much of the apparent effect belongs to the energy deficit rather than to timing, and energy intake is estimated rather than measured. Weight is a surrogate for health and no trial in this literature measured a clinical event. Two of the cited sources have published corrections (Semnani-Azad 2025, and Elortegui Pascual 2023, whose correction fixes a PRISMA record count and states that it does not affect the selected papers), and Lowe 2020 has been corrected twice, so pooled estimates should be read from the corrected versions.

Effect estimate

Versus ad-libitum eating (network meta-analysis of 99 trials, 6,582 adults): alternate-day fasting -3.40 kg (95% CI -4.14 to -2.67, high certainty), whole-day fasting -2.36 kg, continuous energy restriction -2.11 kg, time-restricted eating -1.72 kg. Time-restricted eating versus control (29 randomised trials, 2,485 individuals): -1.37 kg (95% CI -1.99 to -0.75). Alternate-day fasting versus time-restricted eating: -1.69 kg (95% CI -2.49 to -0.88, moderate certainty); versus whole-day fasting: -1.05 kg (95% CI -1.90 to -0.19, moderate certainty).

Sources for this outcome:

Body weight versus matched continuous energy restriction

Finding: Little/no effectCertainty: Moderatestudies found little or no effect

Evidence basis: randomised trials

Once energy intake is matched, no fasting pattern shows a clinically important long-term weight advantage over ordinary calorie restriction.

Population: 1,768 adults across 24 randomised trials in the intermittent-fasting meta-analysis; 2,032 adults with overweight or obesity across 24 randomised trials of at least six months in the longer-term meta-analysis; 100 metabolically healthy adults with obesity over twelve months in the three-arm alternate-day fasting trial; 137 adults with type 2 diabetes over twelve months in the 5:2 noninferiority trial; 139 patients with obesity over twelve months in the calorie-matched time-restricted-eating trial; 116 adults with BMI 27 to 43 over 12 weeks in the TREAT trial.

Research details

Summary

This is the comparison that decides whether fasting adds anything to eating less, and the answer is no. In the 2025 network meta-analysis, alternate-day fasting was the only fasting strategy that beat continuous energy restriction on weight, by 1.29 kg (95% CI -1.99 to -0.59, moderate certainty) - and the same paper describes that difference as trivial, stating that it falls below the 2 kg minimally important difference and suggests somewhat limited clinical impact. In trials lasting 24 weeks or more - 17 of the 99 - only comparisons against ad-libitum diets retained a benefit, so the alternate-day advantage is a short-term phenomenon. Independent syntheses agree: a meta-analysis of 24 randomised trials in 1,768 people found similar weight loss to continuous restriction (mean difference 0.26 kg, 95% CI -0.31 to 0.84, p = 0.37), a meta-analysis of 28 trials in adults found the combined intermittent-energy-restriction diets comparable to continuous restriction (WMD -0.42 kg, 95% CI -0.96 to 0.13, P = 0.132), and the 2026 Cochrane review of intermittent fasting with at least six months of follow-up found it may make little to no difference to percentage weight loss compared with dietary advice (mean difference -0.33%, 95% CI -0.92 to 0.26; low certainty). The two longest single trials agree with the syntheses and cut against the short-term signal: a twelve-month three-arm randomised trial in 100 adults with obesity found alternate-day fasting did not produce superior adherence, weight loss, weight maintenance or cardioprotection compared with daily calorie restriction, and a twelve-month noninferiority trial of a 5:2 pattern against continuous restriction in 137 adults with type 2 diabetes found similar weight change, with the weight and fat-free-mass differences failing their pre-specified equivalence margins. Time-restricted eating behaves the same way: in 139 patients with obesity over twelve months, time-restricted eating plus calorie restriction did not differ from daily calorie restriction alone (net difference -1.8 kg, 95% CI -4.0 to 0.4, P = 0.11), and in 116 adults over 12 weeks the between-group difference against three structured meals was -0.26 kg (95% CI -1.30 to 0.78).

Why this certainty

Rated Moderate and recorded as Little/no effect. It is not lower because the null against matched restriction is the most consistent result in this literature: it appears in the largest network meta-analysis once follow-up reaches 24 weeks, in two independent meta-analyses, in the 2026 Cochrane review and in the two longest single trials, across both whole-day and daily-window patterns. It is not Strong because equivalence cannot be proven by non-significant differences, the confidence intervals are wide enough to hide a small real advantage, comparators are heterogeneous and self-reported, and the one exception - the 1.29 kg alternate-day advantage at short follow-up - is a genuine randomised finding that a reader comparing the two outcomes will meet.

Limitations

Comparators differ across these trials even though all are described as continuous restriction: some prescribed a fixed energy deficit, others gave dietary advice, and some permitted medication changes, so 'matched' is approximate rather than exact. Energy intake is estimated from self-report and adherence is imperfect, which biases against finding a difference in either direction while also making the equivalence claims imprecise. The twelve-month 5:2 trial demonstrated glycaemic equivalence against a 90% confidence interval and failed its pre-specified equivalence margins for weight and fat-free mass, so equivalence on body weight should not be read as equivalence on body composition. Both long trials were small and enrolled specific populations - metabolically healthy adults with obesity, and adults with type 2 diabetes - so neither settles the question in the general population. The Cochrane review graded its evidence low certainty, and none of its included studies reported participant satisfaction or diabetes status.

Effect estimate

Alternate-day fasting versus continuous energy restriction: -1.29 kg (95% CI -1.99 to -0.59, moderate certainty, described by the authors as trivial and below the 2 kg minimally important difference); no fasting strategy differed at 24 weeks or more. Intermittent fasting versus continuous energy restriction: mean difference 0.26 kg (95% CI -0.31 to 0.84; p = 0.37; 24 trials, n = 1,768). Combined intermittent energy restriction versus continuous energy restriction: body weight WMD -0.42 kg (95% CI -0.96 to 0.13; P = 0.132; 28 trials). Intermittent fasting versus dietary advice at six months or more: -0.33% of baseline weight (95% CI -0.92 to 0.26; low certainty; 21 studies, n = 1,430). Time-restricted eating plus calorie restriction versus calorie restriction alone at 12 months: -1.8 kg (95% CI -4.0 to 0.4; P = 0.11; 139 patients). Time-restricted eating versus three structured meals at 12 weeks: -0.26 kg (95% CI -1.30 to 0.78; P = 0.63; 116 adults). 5:2 versus continuous restriction in type 2 diabetes at 12 months: 0.2% HbA1c between-group difference (90% CI -0.2% to 0.5%), meeting equivalence, while the weight difference of -1.8 kg (90% CI -3.7 to 0.07) and the fat-free-mass difference of -0.5 kg (90% CI -1.4 to 0.4) failed their pre-specified equivalence margins.

Sources for this outcome:

Eating-window timing effects specific to time-restricted eating

Finding: Mixed evidenceCertainty: Limitedstudies disagree, or the effect depends on who is studied

Evidence basis: randomised trials

Eating-window timing produces signals that fasting days do not reproduce, but they are inconsistent across syntheses and the largest analysis finds no glycaemic difference at all.

Population: Adults with overweight or obesity, metabolic syndrome, impaired fasting glucose or type 2 diabetes, plus some healthy adults. The largest glycaemic synthesis covered 18 studies and 1,169 subjects; the diabetes-specific meta-analysis covered eight randomised controlled trials and 312 participants; the early-window pooling covered 13 randomised trials and 859 patients; the metabolic-syndrome trial randomised 108 adults (mean age 59 years, BMI 31.22 kg/m2, baseline eating window 14.19 hours).

Research details

Summary

Time-restricted eating asks a question the other patterns do not: not how much you eat but when. The retrieved evidence suggests that when the window falls matters, and it does not establish how much. A meta-analysis of 18 studies in 1,169 subjects found no significant overall effect on fasting glucose (Hedges g = -0.08, 95% CI -0.31 to 0.16) but a reduction in HbA1c (Hedges g = -0.27, 95% CI -0.47 to -0.06) and fasting insulin, and its subgroup analysis found fasting glucose fell only with early time-restricted eating (Hedges g = -0.38, 95% CI -0.62 to -0.14). A meta-analysis of 13 randomised trials in 859 patients found early time-restricted eating reduced body weight (-1.84 kg, 95% CI -2.28 to -1.41) and fat mass (-1.10 kg, 95% CI -1.47 to -0.74). A 2025 network meta-analysis reported time-restricted eating as the most effective strategy against usual diet for lowering fasting plasma glucose (mean difference -3.74 mg/dL, 95% CI -6.01 to -1.46), and a 2025 meta-analysis of eight trials in 312 patients with type 2 diabetes or impaired fasting glucose found reductions in fasting glucose and HbA1c plus increased time in range. Against that, the largest network meta-analysis found no differences between intermittent fasting, continuous energy restriction and ad-libitum diets for HbA1c, and a 2025 meta-analysis of trials in adults with overweight or obesity found no significant effect on fasting plasma glucose or HbA1c. The most useful single randomised result is timing-specific and positive: in 108 adults with metabolic syndrome, a time-restricted eating intervention aligned to circadian timing reduced HbA1c by 0.10% (95% CI -0.19% to -0.003%) against standard of care. So the timing question has a real signal in it, and it is not established for fasting days: no retrieved trial has tested whether shifting an alternate-day or 5:2 pattern earlier in the day changes any of these outcomes.

Why this certainty

Rated Limited and recorded as Mixed evidence. It is not Little/no effect, because the timing subgroup and the metabolic-syndrome trial do show positive effects and the direction for early windows is consistent where it appears. It is not Moderate because the largest and most methodologically rigorous synthesis found no HbA1c difference between strategies, a second large meta-analysis found no effect on either fasting glucose or HbA1c, effect sizes are small, follow-up is short, and the positive results depend on window timing in a way that no synthesis has established as a general property. The outcome is kept separate from the glycaemic comparison against matched restriction so that a timing-specific finding is never read as a property of fasting in general, which is the confusion the consolidation of these two entries could otherwise create.

Limitations

The effect sizes are small and of uncertain clinical importance: the pooled HbA1c reduction in the largest glycaemic meta-analysis was Hedges g = -0.27, and the diabetes-specific pooled reduction was -0.11%. Results conflict across syntheses, with two large analyses finding no effect on HbA1c or fasting glucose, and the units differ across the sources, so the estimates are not directly comparable. Trials are commonly 8 to 12 weeks and some are small and single-centre. Glycaemic benefit appears restricted to early windows, which means protocols are not interchangeable and the entry must not transfer a timing finding from one pattern to another. Pharmacotherapy was permitted in several trials, which can mask or confound dietary effects; reporting of medication changes is incomplete; and several secondary outcomes in individual trials were not adjusted for multiple comparisons. Every outcome is a surrogate, and no trial demonstrates reduced clinical events such as retinopathy, renal disease or cardiovascular events.

Effect estimate

Rovira-Llopis 2024 (18 studies, n = 1,169): HbA1c Hedges g = -0.27 (95% CI -0.47 to -0.06; p = 0.01); fasting glucose Hedges g = -0.08 (95% CI -0.31 to 0.16; p = 0.52); fasting insulin Hedges g = -0.40 (95% CI -0.73 to -0.08; p = 0.01); early time-restricted eating fasting glucose Hedges g = -0.38 (95% CI -0.62 to -0.14; p < 0.01). He 2024 (13 randomised trials, 859 patients): early time-restricted eating versus non-time-restricted eating body weight -1.84 kg (95% CI -2.28 to -1.41), fat mass -1.10 kg (95% CI -1.47 to -0.74), visceral fat area -9.76 cm2 (95% CI -13.76 to -5.75). Nam 2025 (8 randomised controlled trials, 312 participants with type 2 diabetes or impaired fasting glucose): fasting glucose MD -0.74 mmol/L (95% CI -1.13 to -0.36); HbA1c MD -0.11% (95% CI -0.15 to -0.07); time in range +10.51 percentage points (95% CI 6.81 to 14.21). Kibret 2025: time-restricted eating versus usual diet fasting plasma glucose MD -3.74 mg/dL (95% CI -6.01 to -1.46), reported with high certainty. Manoogian 2024 (TIMET, 108 adults with metabolic syndrome): HbA1c -0.10% (95% CI -0.19% to -0.003%) versus standard of care. Semnani-Azad 2025 and Wang 2025: no difference in HbA1c or fasting plasma glucose. Units differ across sources: Kibret 2025 reports mg/dL, Rovira-Llopis 2024 reports Hedges g, and the Nam 2025 time-in-range estimate is a difference in percentage points.

Sources for this outcome:

Glycaemic control, lipids and cardiometabolic risk factors against matched energy restriction

Finding: Little/no effectCertainty: Limitedstudies found little or no effect

Evidence basis: randomised trials

Against matched energy restriction, fasting shows no cardiometabolic advantage: HbA1c did not differ in any comparison, and LDL cholesterol rose in the alternate-day fasting arm of the longest trial.

Population: Adults in the 99 trials of the network meta-analysis, of whom 12 enrolled people with diabetes and 54 people with overweight or obesity; 2,032 adults with overweight or obesity across 24 randomised trials of at least six months; 100 metabolically healthy adults with obesity over twelve months in the alternate-day fasting trial; 137 adults with type 2 diabetes over twelve months in the 5:2 noninferiority trial; adults across 28 trials of intermittent energy restriction in the body-composition synthesis.

Research details

Summary

On risk factors rather than weight, and against energy restriction rather than against unrestricted eating, fasting is close to a null intervention. In the 2025 network meta-analysis, HbA1c showed no difference in any of the ten comparisons, HDL cholesterol showed no difference, and LDL cholesterol differed only for whole-day fasting against an ad-libitum diet (-0.15 mmol/L, 95% CI -0.24 to -0.06); C-reactive protein fell modestly for alternate-day fasting against continuous energy restriction and against whole-day fasting. A meta-analysis of randomised trials lasting at least six months in 2,032 adults found that against a control diet intermittent fasting reduced body weight by 2.84 kg, fat mass by 3.06 kg and waist circumference by 3.85 cm, and lowered diastolic blood pressure by 0.91 mmHg while raising HDL cholesterol by 0.03 mmol/L, but had no effect on insulin, HbA1c, total cholesterol, LDL cholesterol or systolic blood pressure; against continuous energy restriction the only differences were small reductions in fat mass and diastolic pressure. A meta-analysis of 28 trials in adults found no difference from continuous restriction in BMI, glucose, insulin, HOMA-IR, serum lipids, lipoproteins or blood pressure. The most informative single result is a harm signal rather than a benefit: in the twelve-month alternate-day fasting trial, LDL cholesterol rose by 11.5 mg/dL (95% CI 1.9 to 21.1) in the fasting arm relative to daily calorie restriction. For type 2 diabetes, the twelve-month 5:2 noninferiority trial found similar HbA1c reductions to continuous restriction (between-group difference 0.2%, 90% CI -0.2% to 0.5%) and similar weight, while failing its equivalence margins for weight and fat-free mass. The timing-specific glycaemic signals that do exist are recorded separately on the eating-window outcome, because they belong to a narrower question than this one and were not reproduced for fasting days.

Why this certainty

Rated Limited and recorded as Little/no effect. It is not Mixed evidence, because the nulls are nulls rather than disagreements: HbA1c differed in none of ten network comparisons, and the independent syntheses agree that no lipid or glycaemic marker improves against matched restriction. It is not Insufficient direct evidence, because randomised trials did measure these markers against the relevant comparators. It is not Moderate because all outcomes are surrogates, follow-up is short, the pooled analyses use heterogeneous comparators, and the one randomised safety signal that exists points against the intervention.

Limitations

Every outcome here is a surrogate marker rather than a clinical event, and follow-up is short relative to the timescale on which cardiovascular risk changes. The longer-term meta-analysis pooled trials with heterogeneous comparators, so its null results against calorie restriction and its positive results against control diets describe different questions and should not be combined into one narrative. HbA1c equivalence in the type 2 diabetes trial was demonstrated against a 90% confidence interval and the weight and fat-free-mass differences failed their pre-specified equivalence margins, so equivalence on glycaemia should not be read as equivalence on body composition. The LDL rise in the alternate-day fasting arm is a single trial result in metabolically healthy adults with obesity and needs replication before it is treated as a general property of the protocol, but it is a randomised signal and it points the wrong way. No trial in this literature measured cardiovascular events, diabetes incidence, medication escalation or mortality.

Effect estimate

HbA1c: no difference between intermittent fasting, continuous energy restriction and ad-libitum diets in any of the ten network comparisons. Versus control diet at six months or more: weight -2.84 kg, fat mass -3.06 kg, waist circumference -3.85 cm, diastolic blood pressure -0.91 mmHg, HDL +0.03 mmol/L, with no effects on insulin, HbA1c, total cholesterol, LDL cholesterol or systolic blood pressure. Alternate-day fasting versus daily calorie restriction at 12 months: LDL cholesterol +11.5 mg/dL (95% CI 1.9 to 21.1). Whole-day fasting versus ad-libitum diet: LDL cholesterol -0.15 mmol/L (95% CI -0.24 to -0.06). Intermittent energy restriction versus continuous energy restriction across 28 trials: no difference in BMI, glucose, insulin, HOMA-IR, serum lipids, lipoproteins or blood pressure.

Sources for this outcome:

Fat-free mass and lean mass during weight loss

Finding: Mixed evidenceCertainty: Limitedstudies disagree, or the effect depends on who is studied

Evidence basis: randomised trials

Several syntheses report lean-mass loss with fasting, including with daily eating windows, but the largest dedicated analysis did not reach significance and no study links it to harm.

Population: Adults with overweight or obesity and adults with metabolic syndrome. He 2024 pooled 13 randomised controlled trials with 859 patients; Kibret 2025 synthesised 56 studies in adults; Schroor 2024 pooled 28 trials in healthy adults, of which 7 studied time-restricted eating; the TREAT trial measured body composition by dual-energy x-ray absorptiometry in a 50-participant in-person subset (25 time-restricted eating, 25 consistent meal timing), mean age 43.8 years and mean BMI 31.4. Older adults at risk of sarcopenia were not specifically studied.

Research details

Summary

Fasting strategies reduce fat-free mass alongside fat, and this finding is qualified in both directions. Against a control diet, a 2025 network meta-analysis rated it high certainty that time-restricted eating was the most effective intermittent fasting approach at reducing fat-free mass (-0.82 kg, 95% CI -1.46 to -0.17). A meta-analysis of 28 randomised trials in adults found that the intermittent-energy-restriction diets combined reduced fat-free mass more than continuous energy restriction (WMD -0.20 kg, 95% CI -0.39 to -0.01, P = 0.044), and the earlier intermittent-fasting synthesis reported the same pattern, with a pooled reduction of 0.81 kg against a control diet while the comparison against continuous energy restriction was not significant. Against that, the largest dedicated fat-free-mass analysis, a meta-analysis of 13 randomised trials in 859 patients, found that early time-restricted eating did not significantly change fat-free mass overall (-0.56 kg, 95% CI -1.16 to 0.03, P = 0.06, I2 = 59%), and the 16:08 subgroup was the exception rather than the rule. The TREAT trial provides direct randomised corroboration in a substudy, with a significantly greater fall in appendicular lean mass index in the time-restricted arm than in the three-meal control arm (-0.16 kg/m2, 95% CI -0.27 to -0.05, P = 0.005). The important qualification is what kind of finding this is: fat-free mass was a secondary outcome wherever it was measured, no trial was powered for it, the direction depends on the comparator, and not one study measured frailty, falls, strength or functional decline. It is a body-composition signal worth keeping protein intake and resistance training in view for, not a demonstrated harm.

Why this certainty

Rated Limited and recorded as Mixed evidence. Randomised syntheses and one small body-composition substudy report modest reductions in fat-free or lean mass, which is why it is not dismissed; the largest dedicated meta-analysis did not reach significance, the effects vary by comparator, and no study links the changes to frailty or functional decline, which is why it is not graded higher and not labelled a harm. Keeping it as its own outcome rather than a precaution is deliberate: the warning a reader needs is that the lean-mass picture is unresolved rather than that fasting costs muscle.

Limitations

The outcome definitions and comparators differ across syntheses: some compare fasting with unrestricted eating, others with matched calorie restriction, and the direction of the finding depends on that choice, which is why the same literature supports both 'fasting spares nothing' and 'fasting is comparable'. He 2024 reported its primary fat-free-mass result as non-significant with a confidence interval crossing zero and moderate heterogeneity, so it does not by itself establish harm. Fat-free mass was a secondary outcome in the TREAT trial and its body-composition analysis included only the participants who completed in-person testing, so it is underpowered and the between-group weight difference in that subset was itself non-significant. Most trials lasted 12 weeks or less, which is too short to establish whether a change in fat-free mass persists, and none was designed to assess falls, frailty or sarcopenia. No included study reported a clinical endpoint such as injury, loss of function or reduced quality of life attributable to lean-mass change, and no trial enrolled older adults at risk of sarcopenia as a defined population.

Effect estimate

He 2024 (13 randomised controlled trials, 859 patients): fat-free mass -0.56 kg (95% CI -1.16 to 0.03; I2 = 59%; P = 0.06); 16:08 subgroup fat-free mass -0.25 kg (95% CI -0.68 to 0.18; I2 = 0%; P = 0.25). Kibret 2025 (56 studies): time-restricted eating versus usual diet fat-free mass MD -0.82 kg (95% CI -1.46 to -0.17), reported with high certainty. Schroor 2024 (28 trials): combined intermittent energy restriction versus continuous energy restriction fat-free mass WMD -0.20 kg (95% CI -0.39 to -0.01; P = 0.044). TREAT (in-person cohort, n = 50): appendicular lean mass index difference between groups -0.16 kg/m2 (95% CI -0.27 to -0.05; P = 0.005); appendicular lean mass -0.47 kg (95% CI -0.82 to -0.12; P = 0.009).

Sources for this outcome:

What people say

Who discusses this?

Public statements, with positions specific to each source and topic. These views do not determine the scientific evidence assessment.

What the research shows ↑
Show more (2)

Practical steps and precautions

How to try it

  1. Choose the pattern you will actually keep, because the trials that ran for a year found adherence is the binding constraint rather than the schedule: dropout was highest in the alternate-day fasting arm of the twelve-month TREAT trial, and adherence to whole-day fasting fell from about 74% at six weeks to about 22% at a year in the network meta-analysis. Plan the maintenance phase at the same time as the fasting phase, because the weight difference between fasting and ordinary restriction had gone by twelve months.
  2. Whatever the pattern, expect the weight loss to come from eating less overall rather than from the timing. Nearly three-quarters of the comparisons in the network meta-analysis involved a negative energy balance, so tracking what you actually eat matters more than which days you fast or how narrow your window is. If training suffers, recovery worsens or the window makes adequate protein and energy difficult, widen the eating window or choose a different schedule; fasting is not required for a training goal.
  3. If you use a daily eating window, keep it consistent and consider placing it earlier in the day: the timing-specific benefits in the retrieved syntheses come from early windows, and no synthesis established a best window length, so treat 8 or 10 hours as an editorial starting point rather than a validated optimum.
  4. If you take insulin, a sulfonylurea or any glucose-lowering medication, do not start a fasting protocol without discussing it with your prescriber first: fasting increases the rate of hypoglycaemia in type 2 diabetes even when medication is reduced, and the dose adjustments are specific and need monitoring.

Precautions

  • None of these patterns is recommended during pregnancy or breastfeeding, or for children, because safety has not been adequately studied in these populations. They are also not appropriate for anyone who is underweight or undernourished.
  • A history of an eating disorder is a reason not to use a fasting protocol. The retrieved clinical guidance states that intermittent fasting might not be appropriate for people with such a history, and the risk is that a structured fasting schedule provides a rationale for restriction; no outcome data exist either way, so this is a precaution based on clinical judgement rather than on trial evidence.
  • If you have type 2 diabetes and take insulin or a sulfonylurea, the risk of hypoglycaemia is roughly doubled on fasting days and requires protocolised medication reduction and glucose monitoring rather than self-directed fasting. For type 1 diabetes there are insufficient data on efficacy and safety to recommend it at all.
  • Several syntheses report reduced fat-free and lean mass with fasting, including with time-restricted eating. This is a body-composition signal from secondary trial endpoints, not a demonstrated clinical harm, and no included study measured frailty, falls or functional decline. Treat it as a reason to keep protein intake and resistance training in view, and take particular care if you are an older adult who has or is at risk of sarcopenia, because the effect of fasting on muscle mass in that group is unclear.

References

Intermittent fasting strategies and their effects on body weight and other cardiometabolic risk factors: systematic review and network meta-analysis of randomised clinical trials

Semnani-Azad Z, Khan TA, Chiavaroli L, Chen V, Bhatt HA, Chen A, Chiang N, Oguntala J, Kabisch S, Lau DC, Wharton S, Sharma AM, Harris L, Leiter LA, Hill JO, Hu FB, Lean ME, Kahleova H, Rahelic D, Salas-Salvado J, Kendall CW, Sievenpiper JL · BMJ · 2025

Systematic review and network meta-analysis of randomised clinical trials with GRADE certainty assessmentVerified from abstract
Open original source

Intermittent fasting for adults with overweight or obesity.

Garegnani LI, Oltra G, Ivaldi D, Burgos MA, Andrenacci PJ, Rico S, Boyd M, Radler D, Escobar Liquitay CM, Madrid E · The Cochrane database of systematic reviews · 2026

Cochrane systematic review and meta-analysis of randomised controlled trials of at least six months, with GRADE assessmentVerified from abstract
Open original source

Association of Eating Window With Mortality Among US Adults: Insights From a Nationally Representative Study.

Mao Z, Grant H, Kritchevsky SB, Newman AB, Farsijani S · Aging cell · 2025

Prospective observational cohort (NHANES 2003-2018 linked to mortality records), not randomisedVerified from abstract
Open original source

Intermittent fasting activates markers of autophagy in mouse liver, but not muscle from mouse or humans.

Chaudhary R, Liu B, Bensalem J, Sargeant TJ, Page AJ, Wittert GA, Hutchison AT, Heilbronn LK · Nutrition (Burbank, Los Angeles County, Calif.) · 2022

Mechanistic study combining a mouse experiment with a human randomised trial component and muscle biopsiesVerified from abstract
Open original source

Meal Timing and Anthropometric and Metabolic Outcomes: A Systematic Review and Meta-Analysis

Liu HY, Eso AA, Cook N, O'Neill HM, Albarqouni L · JAMA Network Open · 2024

Systematic review and meta-analysis of randomised clinical trials (interventions of 12 weeks or longer)Verified from abstract
Open original source

A meta-analysis comparing the effectiveness of alternate day fasting, the 5:2 diet, and time-restricted eating for weight loss

Elortegui Pascual P, Rolands MR, Eldridge AL, Kassis A, Mainardi F, Le KA, Karagounis LG, Gut P, Varady KA · Obesity (Silver Spring) · 2023

Systematic review and random-effects network meta-analysis of randomised controlled trialsVerified from abstract
Open original source

Does early time-restricted eating reduce body weight and preserve fat-free mass in adults? A systematic review and meta-analysis of randomized controlled trials

He M, Li B, Li M, Gao S · Diabetes & Metabolic Syndrome: Clinical Research & Reviews · 2024

Systematic review and meta-analysis of randomised controlled trialsVerified from abstract
Open original source

Effects of Intermittent Energy Restriction Compared with Those of Continuous Energy Restriction on Body Composition and Cardiometabolic Risk Markers - A Systematic Review and Meta-Analysis of Randomized Controlled Trials in Adults

Schroor MM, Joris PJ, Plat J, Mensink RP · Advances in Nutrition · 2024

Systematic review and meta-analysis of randomised controlled trials in adultsVerified from abstract
Open original source

Effect of Alternate-Day Fasting on Weight Loss, Weight Maintenance, and Cardioprotection Among Metabolically Healthy Obese Adults: A Randomized Clinical Trial.

Trepanowski JF, Kroeger CM, Barnosky A, Klempel MC, Bhutani S, Hoddy KK, Gabel K, Freels S, Rigdon J, Rood J, Ravussin E, Varady KA · JAMA internal medicine · 2017

Twelve-month three-arm randomised clinical trial of alternate-day fasting versus daily calorie restriction versus no-intervention controlVerified from abstract
Open original source

Effect of Intermittent Compared With Continuous Energy Restricted Diet on Glycemic Control in Patients With Type 2 Diabetes: A Randomized Noninferiority Trial.

Carter S, Clifton PM, Keogh JB · JAMA network open · 2018

Twelve-month randomised noninferiority trial of a 5:2 intermittent energy restriction pattern versus continuous energy restriction in type 2 diabetes, with protocolised medication reductionVerified from abstract
Open original source

Calorie Restriction with or without Time-Restricted Eating in Weight Loss

Liu D, Huang Y, Huang C, Yang S, Wei X, Zhang P, Guo D, Lin J, Xu B, Li C, He H, He J, Liu S, Shi L, Xue Y, Zhang H · New England Journal of Medicine · 2022

Randomised controlled trial (12-month, open-label, parallel-group)Verified from abstract
Open original source

Effects of Time-Restricted Eating on Weight Loss and Other Metabolic Parameters in Women and Men With Overweight and Obesity: The TREAT Randomized Clinical Trial

Lowe DA, Wu N, Rohdin-Bibby L, Moore AH, Kelly N, Liu YE, Philip E, Vittinghoff E, Heymsfield SB, Olgin JE, Shepherd JA, Weiss EJ · JAMA Internal Medicine · 2020

Randomised clinical trial (12-week, two-arm, mobile-application based)Verified from full text
Open original source

Circadian alignment of food intake and glycaemic control by time-restricted eating: A systematic review and meta-analysis

Rovira-Llopis S, Luna-Marco C, Perea-Galera L, Banuls C, Morillas C, Victor VM · Reviews in Endocrine and Metabolic Disorders · 2024

Systematic review and meta-analysis of controlled studies of at least 4 weeksVerified from abstract
Open original source

Time-Restricted Eating Improves Glycemic Control in Patients with Type 2 Diabetes: A Meta-Analysis and Systematic Review

Nam T, Oh H, Kim A, Oh Y · International Journal of Molecular Sciences · 2025

Systematic review and meta-analysis of randomised controlled trialsVerified from abstract
Open original source

Time-Restricted Eating in Adults With Metabolic Syndrome: A Randomized Controlled Trial

Manoogian ENC, Wilkinson MJ, O'Neal M, Laing K, Nguyen J, Van D, Rosander A, Pazargadi A, Gutierrez NR, Fleischer JG, Golshan S, Panda S, Taub PR · Annals of Internal Medicine · 2024

Randomised controlled trial (3-month, standard-of-care controlled)Verified from abstract
Open original source

Intermittent Fasting for the Prevention of Cardiovascular Disease Risks: Systematic Review and Network Meta-Analysis

Kibret KT, Peeters A, Tegegne TK, Mesfin YM, Nichols M · Current Nutrition Reports · 2025

Systematic review and random-effects network meta-analysis of randomised trialsVerified from abstract
Open original source

The impact of intermittent fasting on body composition and cardiometabolic outcomes in overweight and obese adults: a systematic review and meta-analysis of randomized controlled trials

Wang B, Wang C, Li H · Nutrition Journal · 2025

Systematic review and meta-analysis of randomised controlled trialsVerified from abstract
Open original source

Longer-term effects of intermittent fasting on body composition and cardiometabolic health in adults with overweight and obesity: A systematic review and meta-analysis.

Khalafi M, Maleki AH, Ehsanifar M, Symonds ME, Rosenkranz SK · Obesity reviews · 2025

Systematic review and meta-analysis of randomised trials with intervention duration of at least six monthsVerified from abstract
Open original source

Intermittent fasting in Type 2 diabetes mellitus and the risk of hypoglycaemia: a randomized controlled trial.

Corley BT, Carroll RW, Hall RM, Weatherall M, Parry-Strong A, Krebs JD · Diabetic medicine · 2018

Randomised controlled trial of intermittent energy restriction versus continuous energy restriction in type 2 diabetes, with continuous glucose monitoringVerified from abstract
Open original source

Background and safety references

These support context and precautions; they are separate from assessed outcome sources.

Intermittent fasting to treat diabetes: time to update clinical practice guidelines.

Varady KA, Chow LS, Peterson CM · The lancet. Diabetes & endocrinology · 2026

Expert Personal View synthesising randomised trials across type 2 diabetes, type 1 diabetes, prediabetes and gestational diabetes, with explicit contraindications and medication-adjustment guidanceVerified from abstractBackground or safety reference
Open original source

Community

Discussion

Have you tried this? What helped or made it difficult?

Reader experiences are separate from the scientific evidence. Explore all discussions

No discussions yet.

Be the first to ask a question or share your experience.

Sign in to vote or join the discussion.

Related habits

Explore other habits in a shared category. Each has its own evidence and precautions.

Editorial history and review method

Revision 4 · Approved for publication · Codex (publication approval explicitly authorized by site owner; category migration)

Prepared and checked with AI. Publication approval does not imply independent expert review. Request a correction.

New entry, added at Standard coverage in the 2026-09-23 pass, covering alternate-day fasting and whole-day energy restriction including 5:2 protocols. It is deliberately distinct from `time-restricted-eating`, which continues to cover daily eating windows; the two entries cross-reference each other and the aliases are disjoint, so a search for 16:8 lands on time-restricted eating and a search for 5:2 or alternate-day fasting lands here. Three outcomes: `body-weight` (Benefit supported, Moderate, randomised, anchored on the 2025 BMJ network meta-analysis of 99 trials and 6,582 adults, with the review's own verdict that the 1.29 kg alternate-day fasting advantage is trivial), `glycaemic-and-cardiometabolic` (Little/no effect, Limited, because HbA1c differed in none of ten comparisons and LDL cholesterol rose in the longest alternate-day fasting trial), and `hard-clinical-outcomes-untested` (Insufficient direct evidence, Not assessed, because no randomised trial has measured events, diabetes incidence or death, and the observational mortality data concern time-restricted eating rather than fasting days). The autophagy claim is addressed in that third outcome from human mechanistic work rather than from animal studies. The precautions cover pregnancy and breastfeeding, children, eating-disorder history, type 1 and type 2 diabetes with hypoglycaemia risk, and older adults at risk of sarcopenia, with the hypoglycaemia contradiction between the guidance paper and the randomised trial reported rather than smoothed over. Second reviewer must check: (a) whether the entry is meaningfully distinct enough from time-restricted-eating or overlaps too far, particularly in the precau-tions and the glycaemic outcome; (b) the trial-name discrepancy for the 5:2-versus-metformin question, which the research pass could not identify and which is therefore not mentioned; (c) whether Standard coverage is right or whether the strength of the weight-loss evidence justifies Full. Scope expanded and consolidated with `time-restricted-eating`, which is retired as a separate entry at the MVP stage; its unique content and source references were migrated here rather than dropped, and its three comparator-implicit outcomes were replaced by five explicit ones plus the retained untested-endpoints outcome. The structure is now body-weight-versus-unrestricted-eating (Mixed, Limited), body-weight-versus-continuous-energy-restriction (Little/no effect, Moderate), eating-window-timing (Mixed, Limited), glycaemic-and-cardiometabolic-versus-energy-restriction (Little/no effect, Limited), lean-mass (Mixed, Limited) and hard-clinical-outcomes-untested (Insufficient direct evidence, Not assessed). Every finding from the two entries is preserved: modest weight loss against unrestricted eating, no clinically important long-term advantage over continuous calorie restriction, a small short-term alternate-day advantage, inconsistent or small glycaemic and cardiometabolic advantages over matched energy restriction, timing-specific effects for eating windows that are not demonstrated for fasting days, qualified lean-mass findings, no randomised evidence on mortality or longevity, and the human-mechanistic answer to the autophagy claim. TRE, 16:8, time-restricted eating, time-restricted feeding, 8-hour and 10-hour eating windows and 'eating window' were added as aliases, and this entry's own previous alias list was preserved in full. The entry states plainly that the patterns are grouped because people search for them together, not because the evidence shows they work the same way, and each outcome names its comparator. The retired entry's review note is reproduced here so its audit history is not lost with the record: "Applied the group-3 audit. body-weight is relabelled Mixed evidence -> Benefit supported, with the headline unchanged, because the divergence is by comparator (time-restricted eating beats ad libitum eating and does not beat matched calorie restriction) rather than a genuine conflict between studies. The limitations now record that none of the 29 trials in the most favourable 2024 pooling was at low overall risk of bias, and that two of the six cited sources plus Lowe 2020 have published errata, with the Elortegui Pascual correction retrieved and quoted in that source record; its canonicalUrl is a PMC link and was not changed. The verdict no longer presents 1.4 to 1.9 kg as one pooled range but attributes about 1.4 kg to the largest pooling against control. glycaemic-control now labels the time-in-range estimate as percentage points and notes that Kibret's fasting glucose is in mg/dL while other estimates are in mmol/L or Hedges g. Second reviewer must re-check: (a) the exact effect of the Elortegui Pascual erratum, which the notice states is a PRISMA record count and does not affect the selected papers; (b) whether the catalogue's rule for Mixed evidence should be tightened for comparator-dependent magnitudes; (c) whether the low-quality HbA1c evidence in other habits should be treated consistently with this relabelling." A second reviewer must check: (a) whether one entry covering daily windows and whole-day fasting is right for readers, or whether the timing outcome should become its own entry again now that the timing-specific evidence has been separated out; (b) whether the body-weight-versus-unrestricted-eating outcome should be Benefit supported rather than Mixed, given the divergence is by comparator rather than between studies; (c) that no TRE-only source was left orphaned by the merge. October 3 training follow-up: Added training-context guidance while retaining the lack of proven lifespan benefit. Independent editorial approval remains pending. Publication release 2026-10-04: site owner authorized a bounded AI audit and release of this exact revision. See docs/release-audit-2026-10-04.md for source-check findings and limits. No independent health-professional or peer review is claimed. Category migration 2026-10-04: removed the separate goals field; categories: Nutrition, Long-term health → Nutrition, Long-term health. Research assessments, sources, guidance, precautions, and evidence review dates are unchanged. This revision awaits publication approval; no new evidence review is claimed. Publication approval 2026-10-04: site owner explicitly accepted all current category-migration revisions for deployment and requested that remaining limitations be retained in docs/content-review-backlog.md. Approval applies to this exact content revision. No new evidence review, independent expert review, or clinical validation is claimed; existing precautions and disclosures remain in effect.