Intermittent fasting (16:8)

What is behind it

In practice, an eating window of eight hours usually means that a meal or evening snacking is dropped. The weight effect therefore works mainly through calories and not through timing as such. That is exactly what the longest high-quality study showed: 139 people with obesity, twelve months, both groups with an identical calorie target, one of them additionally with a window from 8 a.m. to 4 p.m. Both lost considerably; the difference between them was small and not statistically confirmed.

That does not make 16:8 worthless, but it describes its role: a tool that makes a calorie reduction easier to stick with. Anyone who fits the same amount into the window as before notices nothing – in the TREAT study, estimated energy intake did not differ between the groups.

Why the time of day plays a part

Beyond calories, there is a genuine timing component. Glucose tolerance is physiologically poorer in the evening than in the morning: in an eight-day lab protocol, glucose after eating at 8 p.m. was 17 % higher than at 8 a.m., with a 27 % lower early insulin response. This is an effect of the internal clock, independent of the sleep-wake and eating rhythm.

The pattern of the studies fits this. A six-hour window with the last meal before 3 p.m. improved insulin sensitivity, beta-cell response, blood pressure, oxidative stress and appetite – and did so with weight held constant. The study with the latest window and without calorie control is also the one with the weakest result.

The autophagy question

Cellular cleanup is the argument 16:8 is usually sold with, and at the same time its weakest part. Autophagy has barely been measured in fasting humans. Autophagic flux has so far been captured by one randomized trial in blood cells: in 121 people with obesity, the change after 6 months of time-restricted intermittent fasting differed from the control group, but in a post hoc analysis and without a significant increase within the fasting group (Bensalem 2025). In addition, there is a crossover study in 11 overweight adults over four days each: after an early eating window, ketones, cholesterol and the expression of SIRT1 and of the autophagy gene LC3A in whole-blood cells rose in the morning. More gene expression in blood is not the same as measured autophagic flux in muscle, liver or brain.

The flux itself has so far been measured in humans only in the context of the fasting-mimicking diet – in a pilot study with 30 healthy people over eight days, with significant group differences only at individual time points and funded by the manufacturer. That is something different from 16:8. The well-known hour figures come from animal experiments.

What the studies show

Twelve months of a window versus calorie reduction alone

Randomized trial with 139 people with obesity; 118 (84.9 %) completed it. Both groups had the same calorie target (1,500–1,800 kcal for men, 1,200–1,500 kcal for women), one additionally the window 8 a.m.–4 p.m. Result: −8.0 kg with the window versus −6.3 kg without, net difference −1.8 kg (95 % CI −4.0 to 0.4; p = 0.11). The primary endpoint was not met. Waist circumference, body fat, blood pressure and metabolic risk factors showed the same picture.

TREAT: 16:8 without any calorie target

The largest study so far on the popular practice. 116 participants with a BMI of 27–43, twelve weeks; the intervention group ate ad libitum from noon to 8 p.m. Weight loss −0.94 kg (95 % CI −1.68 to −0.20; p = 0.01), but the group difference was only −0.26 kg (95 % CI −1.30 to 0.78; p = 0.63). In the on-site subgroup of 50 people, the appendicular lean mass index fell by −0.16 kg/m² (p = 0.005) to the disadvantage of the time-restricted group.

Early window in prediabetes, without weight loss

Controlled-feeding crossover study in men with prediabetes: a 6-hour eating window with the last meal before 3 p.m. versus a 12-hour window, 5 weeks each. The participants were given enough food for their weight to remain stable. Insulin sensitivity, beta-cell response, blood pressure, oxidative stress and appetite nevertheless improved. This is the best indication that part of the effect does not work through calories.

Where the data stop

Hard endpoints are missing entirely. The umbrella review of 23 meta-analyses and 34 endpoints contains not a single meta-analysis on mortality or cardiovascular events; the high-certainty findings concern surrogate markers exclusively. Then there is study quality: in the meta-analysis of 29 randomized trials, 22 had high concerns regarding risk of bias, and seven more had lesser concerns. The authors call the effect sizes small and of uncertain clinical significance.

How to do it

What the studies actually did: a 7 a.m.–3 p.m. window plus weight-loss counseling over 14 weeks, with adherence of 6.0 days per week – that is the variant with the clearest result. An 8 a.m.–4 p.m. window plus a fixed calorie target over twelve months – that brought −8.0 kg, but no more than the same target without a window. A noon–8 p.m. window without a calorie target over twelve weeks – that brought nothing in the group comparison. The order is therefore set: place the window rather early and do not lose sight of calories.

Safety

Time-restricted eating was well tolerated in the randomized trials: in the twelve-month study there were no substantial differences in adverse events, and in a six-month study in women with PCOS and a very narrow window, no serious adverse events occurred. It is not suitable for pregnant and breastfeeding women, children, underweight people and people with a history of eating disorders. Anyone taking insulin or sulfonylureas changes the basis of their own dosing with a 16-hour window – that belongs in a physician’s hands. And lean mass deserves attention if the window is implemented without strength training and without sufficient protein.

Sources

• Liu D et al., New England Journal of Medicine 2022
• Lowe DA et al., JAMA Internal Medicine 2020 (TREAT)
• Jamshed H et al., JAMA Internal Medicine 2022
• Sutton EF et al., Cell Metabolism 2018
• Jamshed H et al., Nutrients 2019
• Sun ML et al., eClinicalMedicine 2024
• Liu HY et al., JAMA Network Open 2024
• Morris CJ et al., PNAS 2015
• Bensalem J et al., Journal of Physiology 2025