Intermittent fasting: the schedules, the evidence, and what it does not do better

Intermittent fasting is a schedule, not a diet. It tells you when to eat and says nothing at all about what. That distinction sounds pedantic and turns out to explain most of what the research shows.

The intermittent fasting schedules

• 16:8 — all food inside an eight-hour window, most commonly skipping breakfast and eating from midday to eight. By far the most popular form, and the easiest to sustain.
• 5:2 — normal eating for five days, and two non-consecutive days at roughly 500 to 600 calories.
• Alternate-day fasting — a fast day, then an unrestricted day, repeating. Effective in trials and the hardest of the three to keep up.
• Time-restricted eating — the broader term for any fixed window, including gentler versions such as 12:12.
• Extended fasts of 24 hours or more, which sit outside the evidence base most of this article covers and carry their own risks.

What the trials actually show about intermittent fasting for weight loss

Randomised trials comparing intermittent fasting against continuous calorie restriction, matched for total calorie intake, consistently find similar weight loss. This has now been replicated enough — across meta-analyses covering dozens of trials — that it should be treated as settled rather than contested.

The most-cited single study is Lowe and colleagues' TREAT trial in JAMA Internal Medicine: 16:8 produced modest weight loss, no better than the control condition, and a somewhat concerning proportion of the loss came from lean mass.

That finding sounds deflating. It is actually the useful part. If two approaches work equally well, then the choice is not a scientific question at all — it is a practical one, and you get to pick on the basis of which you can sustain. For a substantial number of people, "do not eat before midday" is a far easier rule to follow than "weigh everything and stop at 1,800 calories". A rule you obey beats an optimum you abandon.

Why fasting produces a calorie deficit

Compressing eating into eight hours removes the evening grazing, the late snack and often breakfast. Most people do not fully compensate by eating proportionally more inside the window — the reported spontaneous reduction is typically a few hundred calories a day.

That is the entire mechanism. It is a calorie deficit arrived at by a different route, and the route matters only insofar as it is easier to walk.

Autophagy, and the honest position

Autophagy — the cell's process for breaking down and recycling damaged components — is real, well characterised, and won Yoshinori Ohsumi the Nobel Prize in 2016. Fasting upregulates it in animal models. All of that is true.

What follows from it in humans is much less clear, and the gap is routinely papered over. Measuring autophagy in living people is genuinely difficult; most human data comes from indirect markers in blood or from muscle biopsies that may not represent other tissues. The fasting durations in the rodent work are long relative to a mouse's lifespan and metabolic rate, which makes the translation to human hours unreliable. And no trial has yet demonstrated a clinical outcome in humans attributable to fasting-induced autophagy.

So when you see a specific hour quoted — "autophagy switches on at 16 hours" — that number has not been established in people. It is a plausible extrapolation being presented as a finding.

Where intermittent fasting does look genuinely better

• Simplicity. For people who find rules easier than arithmetic, this is a real advantage and it is the main one.
• Some metabolic markers. Several trials report improvements in insulin sensitivity and blood pressure not fully explained by weight loss, though the effect sizes are modest and not universally replicated.
• Early time-restricted eating specifically. The more interesting recent work suggests that an early window — eating from 8am to 4pm rather than midday to 8pm — may have advantages independent of calories, aligning food intake with circadian rhythms in insulin sensitivity. It is also the version almost nobody wants to do socially.
• Removing late-night eating, which is worth something on its own regardless of the rest.

What breaks a fast

A question asked constantly, and the answer depends on which effect you care about. For the calorie-deficit mechanism, which is the one doing nearly all the work, anything with meaningful calories breaks it. Black coffee, plain tea, water and sparkling water do not. A splash of milk is a handful of calories and matters very little.

For the more speculative autophagy claims, people argue about whether protein specifically interrupts the process. Given that the human autophagy evidence is thin to begin with, optimising a fast around it is optimising around an unknown.

Who should not do intermittent fasting

• Anyone with a history of disordered eating. A restrictive schedule is a poor fit and can be actively harmful — the rigid rules and the extended periods of not eating map uncomfortably closely onto restrictive patterns.
• Pregnancy and breastfeeding.
• Type 1 diabetes, and type 2 on insulin or sulfonylureas, without medical supervision — the hypoglycaemia risk is real and the medication timing has to change with the schedule.
• Underweight adults, and children and adolescents.
• People on medications that must be taken with food.

A note on women and fasting

There is a persistent claim that intermittent fasting is categorically unsuitable for women. The strong version of that is not well supported: most trials include women and find similar results. What is fair to say is that the animal work on reproductive effects used severe restriction, that some women report menstrual disruption anecdotally, and that this deserves more study than it has received. Gentler windows and adequate total intake address most of the concern.

If you want to try it

1. Start with 12:12 and widen gradually. Jumping straight to 16:8 makes the first week unnecessarily unpleasant and is the commonest reason people quit.
2. Keep protein high inside the window. A short window makes it easy to fall short across the day, and the TREAT trial's lean-mass finding is the reason this matters. Two meals of 40 grams of protein beats grazing.
3. Expect the first week to be difficult and the second to be fine. Hunger on a fixed schedule is largely conditioned, and ghrelin adapts to a new eating pattern within days.
4. Do not treat the window as permission. An eight-hour window filled with the same excess is the same excess.
5. Drop it if it makes you miserable. The entire argument for this approach is adherence. If it is harder for you than counting, it has no remaining advantage.