Calories and eating

Week one is the fastest week you will ever have. Almost none of it is fat.

In the best-controlled study of this, seventeen men on a metabolic ward lost 1.6 kg in their first fifteen days of restriction, and 0.2 kg of it was fat.

That is not a reason to be disappointed. The weight is real and the direction is right. But knowing what it was matters, because week two is where people conclude the diet has stopped working, and week two is the first honest number you get.

Published 30 August 2026. This is a summary of published research, not medical or dietary advice.

A worked example

Seven days of an ordinary calorie deficit, weighed at the same time each morning: 82.5 kg to 80.2 kg. A drop of 2.3 kg in a week.

Now the arithmetic. A kilogram of body fat carries roughly 7,700 kcal. To lose 2.3 kg of fat in seven days you would need a deficit of about 2,500 kcal every day, which for most people means eating nothing at all and then some. At a more ordinary 500 kcal a day, seven days is around 3,500 kcal, or something in the region of 400 to 500 g of fat.

So of 2.3 kg on the scale, perhaps half a kilo was fat. The other 1.8 kg was water, glycogen and what was still in transit. All of it genuinely left the body. Almost none of it was what anybody was trying to lose.

Where the rest of it goes

Carbohydrate is stored as glycogen, in muscle and liver, and glycogen is stored wet. Olsson and Saltin loaded nineteen men with carbohydrate over four days in 1970 and measured total body water rising by 2.2 litres and body weight by 2.4 kg. From which they calculated three to four grams of water stored with every gram of glycogen.

That 3:1 figure is now quoted everywhere as though it were a physical constant, and it is worth saying that it is not. It was inferred by division rather than measured directly. A 2015 biopsy study confirmed at least 3 g per gram, but got a ratio nearer 17 to 1 when fluid was freely available, which tells you the number depends on hydration rather than being a property of glycogen. A 2018 study found no measurable water shift at all on depletion. Treat it as a serviceable estimate with real dissent behind it, not a law.

Even at the conventional figure, the ceiling is lower than people assume. Total stores run to a few hundred grams, and clearing 400 to 500 g of glycogen at three parts water is somewhere around 1.6 to 2.0 kg. That is the whole of the water story, and it is a one-time payment. You cannot deplete the same glycogen twice.

And you probably arrived with the stores full

This is the part most explanations miss, and it is well documented. People eat more in the run-up to starting.

Urbszat, Herman and Polivy demonstrated it experimentally in 2002: restrained eaters who were simply told they would be dieting for a week ate significantly more in a taste test immediately afterwards than those who were not. Nothing had been restricted yet. The announcement was enough.

And it shows up at scale. In a treatment trial of 178 adults, 48.9% gained more than 1.15% of their body weight. Averaging 2.5%. Between being screened and turning up to the first session. So a good deal of what comes off in week one went on in the fortnight before it, which is a strange and slightly funny thing to be proud of.

One honest limit: those studies measured windows of weeks, not the two or three days immediately before a Monday start. The effect is real; nobody has isolated the last supper itself.

How much of it is fat, really

It depends almost entirely on how much carbohydrate you cut, and two studies bracket it neatly.

Hall and colleagues put seventeen overweight men on a metabolic ward, fed them a high-carbohydrate diet for four weeks and then switched them to a ketogenic one at the same calories. The first fifteen days produced 1.6 kg of weight loss and 0.2 kg of fat by DXA. Same energy intake, dramatic scale movement, roughly an eighth of it fat.

Yang and Van Itallie ran the older version in 1976, cycling six people through three ten-day diets. An 800-kcal ketogenic diet produced 467 g a day of weight loss that was 61% water and 35% fat. The same 800 calories as a mixed diet produced only 278 g a day, but 60% of it was fat. Six people and ten-day averages rather than a week, so hold it loosely. The direction is the point: the diet that moved the scale faster lost less fat.

Why it slows, and it is not your metabolism

The water is spent. After that the scale reflects fat, which comes off at a few hundred grams a week rather than a few kilos, and no arrangement of food changes that arithmetic much.

As for the resistance that builds: the popular answer is a collapsing metabolism, and the measured answer is appetite. In a 52-week trial of 153 people where the deficit was created covertly, so nobody was consciously dieting against it. Weight loss drove hunger up by roughly 100 kcal a day for every kilogram lost. That is more than three times the size of the drop in energy expenditure over the same period.

Which is a more useful thing to know, because it is a different problem. The metabolic adaptation story has its own page, including why it is real and much smaller than advertised.

The bit everyone gets wrong about the slowdown

The standard telling is that the big first week sets an expectation, week two disappoints, and people quit. It is intuitive, it is repeated constantly, and the evidence does not support it.

Nackers and colleagues sorted 262 women by how fast they lost in the first month. At eighteen months the fastest group was down 10.9 kg and the slowest 3.7 kg, and there was no difference in regain between them. Fast starters did better, not worse.

What predicts quitting is the opposite: getting nothing. Among 5,274 people on a commercial programme, those who had lost nothing at all by the end of month one were 86% more likely to drop out early. And when researchers tested the disappointment mechanism directly in 180 adults, failing to meet weight-loss expectations lowered people’s satisfaction but had no relationship to attrition or regain.

So the fast start is good news, and the risk is narrower than the folklore suggests: not that you will be disappointed, but that you will read week two as evidence the diet stopped working, when week two is simply the first week that was ever telling you the truth.

That last sentence is an inference from the mechanism rather than a finding. Nobody has tested it. The findings are the two above it, and they both point the same way.

What to do with the number

Do not set your expectations from week one. Take the average from week three onwards, once the water has gone and there is nothing left in the reading but fat and noise.

Expect the scale to wander, and do not read the wander. Across 4,657 daily weigh-ins from 80 adults, weight followed a weekly rhythm. Highest on Sunday and Monday, falling through the week. The authors’ own conclusion was that this variation should be treated as normal rather than as a sign of gaining. Worth adding that the confident figures people quote for daily fluctuation do not appear to have a source; the pattern is documented, the amount is not.

And expect the arithmetic to overpromise. The 3,500 calories per pound rule was tested against seven studies with confined or objectively measured intake and overestimated real loss in all of them. The model behind the NIH Body Weight Planner puts the half-time of the body’s response at about a year. The curve flattens long before the weight settles, and that is the expected shape rather than a fault.

Fuel8 is on the App Store.

A food diary for iPhone. It tracks protein, carbs and fat alongside calories, shows a weight trend rather than a daily verdict, never adds exercise calories back, and has no streaks to break.

Download on the App Store

Or leave your name and email for new recipes and what we add next. No more than that.

This form is the website, not the app. Your name and email go to Grow More Services L.L.C-FZ, the company that builds Fuel8, and are stored in Mailchimp so we can email you about new recipes and what we add next. This list is separate from your optional Fuel8 account. Unsubscribe in one click, any time.

Common questions

Why do you lose so much weight in the first week of a diet?

Because most of it is not fat. Carbohydrate is stored as glycogen, and glycogen is stored wet. The conventional estimate is 3 to 4 grams of water per gram of glycogen (Olsson & Saltin, 1970). Cut carbohydrate and energy and those stores draw down over a few days, taking their water with them. In a metabolic-ward study where 17 men switched to an isocaloric ketogenic diet, the first 15 days produced 1.6 kg of weight loss and 0.2 kg of fat (Hall et al., 2016).

How much of week one is actually fat?

It depends almost entirely on how much carbohydrate you cut. In Hall's 2016 ward study, roughly an eighth of the first fortnight's weight change was fat. Yang and Van Itallie (1976) found that over ten days an 800-kcal ketogenic diet produced weight loss that was 61% water and 35% fat, while the same calories in a mixed diet produced less weight loss but 60% of it fat. Move less carbohydrate and the scale moves less while the fat loss stays similar.

Is it true that people overeat before starting a diet?

Yes, and it has been measured both ways. Urbszat, Herman and Polivy (2002) found that restrained eaters simply told they would begin a week-long diet ate significantly more in a subsequent taste test than those who were not. In a treatment trial of 178 adults, 48.9% gained more than 1.15% of their body weight. An average of 2.5%. Between the screening visit and the first session (Tronieri et al., 2018).

Why does weight loss slow down after the first week or two?

Partly because the water is a one-time payment. You cannot deplete the same glycogen twice. Beyond that, the dominant effect is appetite rather than metabolism: in a 52-week trial of 153 people, weight loss drove appetite up by roughly 100 kcal a day for every kilogram lost, more than three times the size of the accompanying drop in energy expenditure (Polidori et al., 2016).

Does the fast start mean I will quit when it slows?

The evidence says the opposite, which surprises people. Nackers and colleagues (2010) followed 262 women and found the fastest first-month losers were down 10.9 kg at 18 months against 3.7 kg for the slowest, with no difference in regain between them. And it is the absence of early loss that predicts dropping out: in 5,274 commercial-programme participants, losing nothing at all by the end of month one raised the odds of early dropout by 86% (Alexander et al., 2018).

So what rate should I actually expect after week one?

Slower than week one, and slower than the arithmetic suggests. The familiar 3,500 calories per pound rule substantially overestimates real loss. Thomas and colleagues (2013) tested it against seven studies with confined or objectively measured intake and it failed in all of them. The model behind the NIH Body Weight Planner puts the half-time of the body's response to a change in intake at around a year (Hall et al., 2011), so the curve flattens long before the weight settles.

More in this series: how long the average diet lasts, what happens when you eat too little, and eating out without wrecking it.

Sources

This page summarises published research and is not medical or dietary advice. Studies report group averages and individual results vary. The worked example is one person’s seven days and is used to show the arithmetic, not as a typical result. If you have a medical condition, are pregnant, or have any history of disordered eating, speak to a qualified professional before changing how you eat.