Intermittent fasting attracted enormous attention on the basis of animal research and mechanistic reasoning. The human trials have now accumulated and are worth looking at directly.
The variants
Time-restricted eating confines intake to a window each day, commonly eight hours.
Alternate day approaches alternate normal eating with fasting or very low intake.
Periodic approaches involve two low-intake days weekly with normal eating otherwise.
These differ enough that evidence for one does not transfer to the others, and they are frequently discussed as though they were one thing.
The weight loss question
The most studied outcome and the clearest answer.
Trials comparing intermittent approaches with continuous calorie restriction, matched for total energy intake, generally find similar weight loss.
Which suggests the mechanism is reduced total intake rather than anything specific to the timing.
Fasting windows do reduce intake for many people, simply because there is less time to eat, and that is a real practical effect even if it is not a metabolic one.
The metabolic claims
Where the animal research generated the most excitement.
Rodent studies found effects on lifespan, metabolic markers and cellular repair processes.
Human trials have found improvements in some metabolic markers, and controlled comparisons with matched calorie restriction generally attribute them to weight loss rather than to fasting itself.
Some trials with tightly controlled feeding, where weight was held constant, have found effects on insulin sensitivity and blood pressure, which is the most interesting evidence and comes from small studies.
Autophagy
The most frequently cited mechanism and the least well supported in humans.
Autophagy is a genuine cellular process of recycling damaged components, and it is upregulated by nutrient deprivation.
Most of the evidence is from cell and animal studies.
Measuring autophagy in living humans is technically difficult, which means the fasting durations claimed to trigger it in humans are largely extrapolated rather than measured.
The confidence with which specific hour thresholds are stated online has no corresponding evidence.
Muscle mass
A genuine concern with some support.
Weight loss by any method includes some lean mass, and some trials of fasting approaches have found greater lean mass loss than continuous restriction.
Adequate protein intake and resistance training mitigate this, which applies to any weight loss approach.
Compressing eating windows can make adequate protein intake harder, which is the practical mechanism.
Adherence
The variable that determines whether any dietary approach works, and the evidence is mixed.
Some people find time restriction considerably easier than counting calories, because the rule is simple and requires no measurement.
Others find it unsustainable, and dropout rates in trials have sometimes been higher than in comparison conditions.
Which suggests it works well for the people it suits, which is true of every dietary approach and is the honest summary of the field.
Who should not
People with a history of disordered eating, since restrictive rules around eating can precipitate or maintain problems.
People with diabetes on medication affecting blood glucose, where fasting risks hypoglycaemia without medication adjustment.
Pregnant and breastfeeding women.
Children and adolescents.
And anyone underweight or with significant medical conditions.
Which is a substantial list, and it is why this warrants a conversation with a doctor rather than being adopted on the basis of reading.
The summary
It is one reasonable way to reduce intake, no better and no worse than others for weight, with some interesting metabolic questions unresolved.
That is a considerably duller conclusion than the coverage, and it is what the trials support.
Time-restricted eating specifically
The variant with the most recent research and the most interesting open question.
Studies have examined whether the timing of the eating window matters, comparing early windows with late ones.
Some evidence favours earlier windows on metabolic markers, consistent with circadian influences on glucose handling.
Which would suggest that skipping dinner is metabolically preferable to skipping breakfast, and the human evidence is limited and the studies small.
It is one of the more genuinely open questions in the area.
Breakfast
Long-standing claims about breakfast being essential have not held up well in controlled trials.
Observational associations between breakfast skipping and poor outcomes appear substantially confounded, since breakfast skippers differ in many other respects.
Trials assigning people to eat or skip breakfast have generally found little difference in weight outcomes.
Which is a good example of an observational finding that became conventional wisdom before being tested.
Practical difficulties
Social eating, family meals and work schedules all interact badly with fixed windows.
Which is a real cost that rarely appears in trial data and frequently determines whether someone continues.
Flexible application, where the window moves when circumstances require, appears more sustainable than strict adherence.
Medication timing
Several medications require food, and some require specific timing relative to meals.
Which interacts directly with a restricted eating window and is a practical reason to discuss any fasting approach with a pharmacist.
This applies particularly to diabetes medication, where the interaction is potentially serious.