Skip to content
Fasting Index

2 days a week · beginner

The 5:2 diet

Two days a week at roughly 500 to 600 calories, five days of ordinary eating. The two restricted days are not true fasts. No trial that compared it directly against eating a little less every day found a significant weight difference either way, and the one that asked about hunger found more of it on this schedule.

Who should not do a 5:2 fast

  • Anyone taking insulin or a sulfonylurea, unless the schedule and the doses are managed with the prescribing clinician. In the 12-month diabetes trial, medication was reduced at baseline under a written protocol and 35 percent of the people on those drugs still had a hypoglycaemic or hyperglycaemic event in the first two weeks.
  • Anyone who is pregnant or breastfeeding.
  • Anyone with a current or past eating disorder. A recurring rule-bound low intake day is a known relapse trigger, and the weekly rhythm can normalise restrict and rebound cycling.
  • Children and adolescents.
  • Anyone underweight, or losing weight without intending to.
  • Anyone on medication that must be taken with food, or whose dose depends on food intake. A 500 calorie day is where that becomes a problem.

This is not medical advice and no clinician has reviewed it. If any of the above applies to you, speak to a doctor before fasting rather than after.

A schematic timeline of a fast from 0 to 72 hours. A tone band along the top darkens as the body moves from fed to deeply fasted. Beneath it, liver glycogen is drawn as a stipple field that is largely spent within the first 24 hours. Below that, three traces on independent scales: glucose falls then levels, insulin falls steeply and stays low, ketones rise from about 12 hours and climb through the second and third day. The pattern is well described but individuals vary widely, and the horizontal scale breaks at 24 hours.
The hours of a fast. Above, the phases as a tone that darkens as the body moves from fed to deeply fasted. In the middle, liver glycogen as stipple, largely spent inside the first day. Below, three traces on independent scales: glucose, insulin, ketones. A schematic of a well-described pattern. Individuals vary widely, and the scale breaks at 24 hours.

What is actually happening

The 5:2 asks for two non-consecutive days a week at about a quarter of normal energy intake, with ordinary eating on the other five. On a day of 500 to 600 calories insulin stays lower for longer than usual and fat oxidation rises, but food is still arriving, so the body does not reach the state a full day without food produces. Liver glycogen is drawn down rather than emptied. What does the work, when the diet works, is arithmetic rather than physiology. Two days at roughly 25 percent of needs sets a weekly deficit of about 25 percent, which is the same deficit the comparison diets in these trials were built around. The original trial made that explicit by prescribing an identical 25 percent restriction to both arms, one spread across seven days and one concentrated into two. That is the reason it is worth reading. It holds the deficit equal and lets the schedule answer for itself, and when the deficit is held equal the schedule stops explaining much.

What the evidence shows

  • In the trial that established the schedule, 107 premenopausal women with overweight or obesity were randomised to the same 25 percent energy restriction taken either on two days a week or spread across all seven. After six months the weight loss did not differ: 6.4 kg on the two-day schedule against 5.6 kg on the daily one, P = 0.4.
  • That same trial found reductions in fasting insulin and insulin resistance were greater on the intermittent schedule than the continuous one, a between-group difference of 1.2 μU/ml for fasting insulin at P = 0.04. The paper describes the reductions in both groups as modest.
  • In a 12-month randomised trial of 137 adults with type 2 diabetes, two restricted days a week reduced HbA1c equivalently to daily calorie restriction, a between-group difference of 0.2 percent with a 90 percent confidence interval of −0.2 to 0.5 percent, meeting the trial's prespecified equivalence margin.
  • Weight loss in that diabetes trial was 6.8 kg on the intermittent schedule against 5.0 kg on the continuous one, P = 0.25, and the difference did not meet the equivalence criterion the trial set for itself. The trial therefore could not conclude the two diets were the same on weight, in either direction.
  • In a one-year randomised trial of 112 adults with abdominal obesity, participants on two restricted days a week reported significantly higher hunger than those restricting every day, 4.7 against 3.6 on a subjective rating scale, P = 0.002, while losing a similar amount of weight, 8.0 kg against 9.0 kg, P = 0.6.
  • In a 2024 meta-analysis of 28 randomised trials of intermittent energy restriction, body mass index fell less on the 5:2 diet than on continuous calorie restriction. Pooling all three intermittent schedules, the difference in body weight against continuous restriction was 0.42 kg and was not statistically significant, P = 0.132.
  • In the 12-month diabetes trial, hypoglycaemic or hyperglycaemic events in the first two weeks affected 35 percent of the participants taking sulfonylureas or insulin, 16 of 46, and this happened after their medication had already been reduced at baseline under a written protocol.

What the evidence does not support

  • The evidence does not support the 5:2 as better than eating fewer calories every day. Three randomised trials running six months, twelve months and twelve months found no significant weight difference against continuous restriction, and the 2024 meta-analysis found body mass index fell less on the 5:2 than on continuous restriction.
  • The 5:2 is described everywhere as fasting, and on the schedule these trials tested it is not one. The restricted days carry 400 to 600 calories, so findings from 24-hour or longer fasts, including anything about deep ketosis or autophagy, do not transfer to it.
  • The insulin results from the original trial are the strongest positive finding on this page, and they come from a single six-month trial in premenopausal women, with reductions the authors themselves call modest. The other trials on this page did not reproduce them.
  • Nothing here establishes outcomes beyond one year. The longest trials cited ran 12 months, and the pooled analysis includes trials as short as two weeks.
  • The higher hunger on this schedule was measured on a subjective rating scale in one trial. It is a real and statistically significant difference, and it does not establish how many people give the diet up because of it.
  • The diabetes trial reduced hypoglycaemia-causing medication at the start under a medication management protocol. Its safety record cannot be read as the safety record of someone starting this at home on the same drugs without that step.

How it is run

  • Two non-consecutive days a week at a reduced intake. The published trials used 400 to 600 calories depending on sex, or 500 to 600 calories for everyone, so there is no single number the evidence points at.
  • The other five days are ordinary eating rather than compensatory eating. No trial licensed a larger intake on the five days, and the weekly deficit is the entire mechanism.
  • Protein and micronutrients are the practical constraint on a restricted day. Five hundred calories is not much room, and this is the part of the diet that tends to be planned last.
  • Expect the restricted days to feel harder than the calendar suggests. The one-year trial that asked found hunger significantly higher on this schedule than on daily restriction.

What breaks a 5:2 fast

The rules differ by protocol. Check any food, drink or supplement against this protocol below.

See the full index

Sources

No clinician reviews this site. Sources on this page were verified against PubMed by the author, who is not a doctor or a dietitian. What that means.

Last updated 2026-08-16.

The phases of a fast covers the physiology behind every protocol on this site, hour by hour.