The first two weeks of intermittent fasting: what to expect

Day one is the worst day, and it is mostly caffeine

Fasting headache is real and well documented. In a study of 370 hospital staff, 39 percent of those who fasted for 25 hours developed a headache, against 7 percent of those who did not. The risk rose with the length of the fast and was much higher in people with a history of headaches.

The reassuring part is the time course. In a 2024 study, headache affected 33 percent of people on day one and 11 percent by day seven. A separate trial recorded 46 percent on day one in its placebo group, with the incidence dropping over the following days, and by the second week the drug being tested no longer made any difference.

The likely cause is more actionable than the fasting itself. Caffeine withdrawal headache affected 55 percent of caffeine consumers on the first day of fasting, and the risk rose with how much caffeine they normally drank. One study found caffeine withdrawal to be the most important factor; another concluded it did not appear influential, so the field is not unanimous. But if you drink coffee daily and are about to change when you eat, that is the variable worth managing first.

The uncomfortable finding about hunger

You will have been told that hunger fades after a week or two as your body adapts and ghrelin resets. The controlled data does not support that, at least not for every approach.

In a 22-day study of sixteen non-obese adults doing alternate-day fasting, hunger rose on day one and remained elevated throughout. Ghrelin did not change significantly. The authors concluded this showed the unlikelihood of continuing that diet for extended periods. A separate study of six weeks of morning fasting found little adaptation in appetite. A 12-week study of intermittent energy restriction found ghrelin went up rather than down, though it reported no differences between its groups.

There is evidence pointing the other way, and it is worth having. Four days of early time-restricted eating lowered mean ghrelin and made hunger, in the researchers words, more even-keeled.

The honest reading is that it depends on which approach you choose. Daily time-restricted eating may become easier. Alternate-day fasting, in the trials, did not.

The "fasting flu" is mostly not a thing

This is the good news and it is better than the internet suggests. A meta-analysis of 15 randomised trials covering 1,365 adults found no excess fatigue in fasting groups compared with controls, no excess headache across the trial period, and no excess dropout. There was a small non-significant signal for dizziness, and it came from the subgroup doing fasting without an early eating window. Everyone in these trials had overweight or obesity, so this is not a general population result.

Sleep and mood held up too. A trial of 197 people using accelerometers to measure sleep objectively, rather than asking, found that 12 weeks of eight-hour time-restricted eating had no effect on sleep, depression, anxiety, stress or quality of life.

The Ramadan literature does show about an hour less sleep, but that involves eating before dawn and a shifted social schedule, so it says little about someone shifting breakfast to noon.

Will you stick with it?

More often than the reputation suggests. In a one-year trial, dropout was 38 percent for alternate-day fasting against 29 percent for daily calorie restriction and 26 percent for the control group. Higher, but not dramatically, and the pooled difference across trials was one percent and not statistically significant.

That same trial found something more useful than the dropout figure: people doing alternate-day fasting ate more than prescribed on fasting days and less than prescribed on eating days. The protocol drifted toward the middle. If you are choosing between approaches, a daily window that you follow is worth more than an aggressive schedule you quietly average out.

What predicts dropping out is not something I could find a clear answer on. What the trials do suggest is that fit with your life matters more than the schedule you pick.

Things people recommend that have no evidence

Two pieces of standard advice have nothing behind them, and it is worth knowing which.

Electrolytes for early fasting symptoms. No trial has tested this in intermittent fasting. The nearest study looked at electrolytes during severe energy restriction and exercise in heat, which is a different question. For a 16-hour daily fast with normal meals, this is a supplement being sold on a premise nobody has tested. That scoping matters: on much longer fasts, and for anyone taking a diuretic or living with kidney problems, fluid and salt balance stops being a supplement question and becomes a medical one.

Ramping your window gradually. Zero trials. It is intuitively sensible, it is what most people are told, and it has never been compared against simply starting.

That does not make either harmful. It makes them untested, which is different from wrong, and worth distinguishing from the caffeine finding, which has real numbers.

Week one, in order

Deal with caffeine before you deal with anything else. It is the one variable with quantified evidence behind it.

Start with a window you would keep on a bad day. The trial evidence says people drift toward the middle of whatever they set, so set something near where you will actually land.

Judge nothing by the scale for a fortnight. The first two weeks move a couple of kilograms of water and glycogen in either direction, which tells you nothing about fat.

And if hunger has not faded by week two, you are not doing it wrong. In several controlled studies, it did not fade. That is a reason to change the schedule, not to conclude you lack discipline, and widening your window or stopping altogether are both legitimate answers. If hunger is dominating your day, or you find yourself thinking about food constantly, eat more rather than pushing harder. If food has been a difficult subject for you in the past, this is worth raising with a doctor rather than solving with a tighter window.