Fasting and plain calorie cutting come out almost level

Intermittent fasting has been sold as both a breakthrough and a fad. A review published in The BMJ in June 2025 pooled 99 randomised trials, which allocate people to one approach or another by chance, covering 6,582 adults, and the answer it produces is neither. Most fasting schedules lost about as much weight as simply eating less every day. One pattern, alternate day fasting, did better, and the margin was 1.29 kg.

That number is worth sitting with, because it is smaller than the conversation around fasting would suggest, and it is also not nothing. Against an unrestricted diet, alternate day fasting produced 3.40 kg of weight loss. Ordinary daily calorie restriction produced 2.11 kg. Time restricted eating, the popular version where you compress meals into a window of six or eight hours, produced 1.72 kg, and showed no advantage over daily calorie cutting at all.

How we know

This is a network meta-analysis, which is worth explaining because it is doing something ordinary reviews cannot. A standard meta-analysis pools trials that compared the same two things. A network analysis links trials that share a common comparator, so if many trials compared fasting against an unrestricted diet and many others compared daily calorie restriction against the same thing, the method can estimate how fasting and calorie restriction compare even where few trials tested them against each other directly.

That is the method's strength and its weakness at once. It answers questions nobody ran a trial for. It also means the answer is partly inferred rather than observed, and in this case that matters: only three of the 99 trials compared one fasting strategy directly against another. The authors flag this themselves as a limitation of indirectness.

The trials covered 720 healthy adults and 5,862 with an existing health condition, most commonly overweight or obesity, type 2 diabetes or metabolic syndrome. Follow up ran to a median of 12 weeks, with a range from 3 to 52. The authors graded certainty using GRADE, a standard system for rating how much confidence an estimate deserves, and most comparisons came out low to moderate rather than high.

The two comparisons that earned high certainty were alternate day fasting and whole day fasting, both against an unrestricted diet, which is the least surprising finding in the paper: going without food for whole days at a time produces weight loss compared with eating whatever you like. The interesting comparison, alternate day fasting against daily calorie restriction, came out at moderate certainty and 1.29 kg, with a confidence interval, the range within which the true value can reasonably be expected to lie, running from 0.59 kg to 1.99 kg.

What the four patterns actually meant

The labels matter, because they describe quite different weeks. Alternate day fasting means eating normally one day and taking little or nothing the next, repeating. Whole day fasting means one or two full fast days in an otherwise normal week, the pattern popularised as 5:2. Time restricted eating means eating every day, but only inside a fixed window, commonly eight hours. Continuous energy restriction is the unglamorous comparator: eat a bit less than you burn, every day, with no timing rules at all.

Read that list against the results and a pattern appears. The approach that produced the most weight loss is also the one that removes the most food from the week, and the approach that produced no advantage over daily calorie cutting is the one that changes when you eat without necessarily changing how much. That is not proof of a mechanism, since these trials were not designed to isolate one. It is a consistency worth noticing.

What happened to the blood work

Many people fast for reasons that have nothing to do with the scales. They are after better cholesterol, steadier blood sugar, lower insulin. The trials measured these, and the headlines largely skipped them.

The short version is that almost nothing moved. There were no differences in HbA1c, the standard measure of average blood sugar over the preceding months, across any comparison. There were no differences in HDL cholesterol either, the fraction that carries cholesterol away to the liver. The one signal that did appear ran in an unwelcome direction: against time restricted eating, whole day fasting showed slightly higher LDL cholesterol, the fraction that builds fatty plaque in the arteries as it rises, by 0.11 mmol/L, along with small increases in non-HDL and total cholesterol. Certainty on those lipid findings was rated low, which means they could as easily be noise as signal.

So a reader who fasts to improve their metabolic markers should read this review carefully rather than triumphantly. It does not show fasting improving those markers more than ordinary calorie restriction does. It mostly shows the two approaches arriving in the same place.

Why it matters

The practical implication is not that fasting fails. It is that the fasting schedule is doing less work than the calorie deficit underneath it. Across 99 trials, four quite different eating patterns converged on a similar answer, and the pattern that did best is also the hardest to sustain, which the trials were too short to test.

That reframes the question. If several approaches produce roughly the same result over three months, the useful question is not which one is optimal but which one a particular person can actually keep doing. That is a question about individual circumstance, and no meta-analysis can answer it.

It also puts a boundary around what has been established. Twelve weeks is long enough to move a number on a scale and short enough that it says nothing about the thing most people actually want, which is a change that lasts. On that, the evidence is close to silent, because the trials that would answer it have largely not been run.