How Many Calories Are in a Pound of Fat? The 3,500 Rule, and Where It Breaks
Every result agrees on the number: 3,500 calories in a pound of body fat, 7,700 in a kilogram. The number is real and it has an author. It is also an assumption about tissue composition dressed up as a physical constant, and when you use it to plan, it fails in one direction only. It always tells you that you should have lost more.
The short answer
A pound of body fat is counted as 3,500 calories and a kilogram as 7,700 — Wishnofsky's 1958 estimate, not a measurement. Because expenditure falls as you get lighter, that rule overpredicts loss, so PlateLens prompts you to recompute your target when your weight has moved.
In metric literature the same rule is written 32.2 MJ/kg, which is where the 7,700 comes from (Kevin Hall, International Journal of Obesity, 2008). Both forms describe the same claim: that a pound or a kilogram of body weight, lost or gained, is worth that much stored energy. It is a reasonable working estimate over short horizons. It is not a law, and the rest of this article is about the difference.
Where 3,500 came from
One paper. Max Wishnofsky, writing in the American Journal of Clinical Nutrition in 1958, worked out that with adequate protein intake, the energy equivalent of one pound of body weight gained or lost is 3,500 calories. That is the entire provenance of a number now printed on the side of treadmills.
The detail almost nobody repeats is that Wishnofsky attached a caveat in the original: the value is much less during fasting, or during refeeding after a fast. The rule arrived qualified and lost its qualification somewhere in seventy years of retelling.
It is not folk wisdom, either. The arithmetic is embedded in clinical guidance. The 1998 NHLBI expert panel guideline recommended a deficit of 500–1,000 kcal per day to produce about 1–2 lb of weight loss per week. Multiply out the low end: 500 × 7 = 3,500, one pound. The rule is not sitting beside the guidelines; it is inside them.
Why it isn't 9 calories a gram times 454
The obvious objection is that fat holds 9 calories per gram, so a pound of it should be worth about 4,080. Run it:
Dietary fat: 9 kcal/g (37 kJ/g) — the Atwater general factor, alongside 4 kcal/g for protein and carbohydrate and 7 for alcohol (FAO Food and Nutrition Paper 77).
A pound of pure lipid: 453.6 g × 9 kcal/g ≈ 4,080 kcal. A kilogram of pure lipid: 9,000 kcal.
Wishnofsky's pound: 3,500 kcal — about 86% of the pure-lipid figure, consistent with his assumption that the tissue lost is roughly 87% fat.
So the 4,080 is not wrong; it is answering a different question. Two things separate it from 3,500.
First, 9 kcal/g is a food number. The FAO publishes it as the metabolisable energy you get from digesting dietary fat. It describes what arrives, not what is stored, and it cannot be applied straight to adipose tissue, which also carries water and protein.
Second, the weight you lose is not adipose tissue. It is a mixture of fat mass and fat-free mass, and fat mass carries roughly nine times the energy density of fat-free mass (Hall, British Journal of Nutrition, 2007). The proportion depends on how much body fat you started with and on which direction you are moving. That makes the energy content of a kilogram of lost body weight a composition-dependent quantity, not a constant.
Hall's 2008 modelling puts a number on how much that matters. Using a modified Forbes equation, 32.2 MJ/kg approximately matches only people whose initial body fat is above about 30 kg. For leaner people it overestimates the deficit required per unit of weight lost. The single figure that everyone quotes is a decent fit for one part of the population and a poor one elsewhere.
Where the rule breaks: it assumes you never get lighter
This is the failure that matters, and it is not subtle. Hall and Chow, writing in the International Journal of Obesity in 2013, identify the rule's most serious error as treating energy balance as a static quantity when it is dynamic. A static rule predicts weight loss continuing in a straight line with no plateau, which is not what bodies do, and the consequence they state plainly is that it “leads to overestimation of weight loss in individuals and populations.”
The mechanism fits in one sentence: a smaller body costs less to run. Every pound you lose lowers your expenditure a little, so the deficit you carefully set up shrinks underneath you while your intake stays exactly where you put it. The rule never sees this happen, because it has no term for it.
Thomas and colleagues checked it against reality in the same 2013 journal issue, comparing rule-predicted against actual weight loss in seven experiments conducted under total supervision or with objectively measured energy intake. Their conclusion: the 3,500-kcal rule grossly overestimates actual weight loss, even used as a convenient estimate.
The drift, in numbers
| Sustained daily deficit | 3,500 rule says, 12 weeks | 3,500 rule says, 1 year | Where the dynamic model settles |
|---|---|---|---|
| 250 kcal/day | 6 lb (2.7 kg) | 26 lb (11.9 kg) | ~25 lb (~10 kg) |
| 500 kcal/day | 12 lb (5.5 kg) | 52 lb (23.7 kg) | ~50 lb (~21 kg) |
| 750 kcal/day | 18 lb (8.2 kg) | 78 lb (35.6 kg) | ~75 lb (~31 kg) |
Read the middle row across, because it is the one the research speaks to directly. A 500 kcal/day reduction is almost exactly the 2 MJ/day case Hall's team modelled in The Lancet in 2011, and for that reduction the static rule predicted roughly 100% more first-year weight loss than the validated model did. Not 10% out. Double.
At 12 weeks the gap is survivable, which is precisely why the rule feels trustworthy at first. Extend the same straight line to a year and, at that middle-row deficit, it is promising roughly twice the loss the model does. Two things about that last column: it assumes the reduction is sustained permanently, which almost nobody does, and it is there to show the shape of the curve rather than to be a target. Note also that the two published roundings of the model's rule of thumb, 10 kcal/day per pound and 24 per kilogram, do not convert into each other exactly. That is what a rule of thumb looks like when it is honest about being one.
The first three weeks run the other way
The rule is not uniformly pessimistic. Early on it is too conservative, which is the other half of why it feels like a betrayal later.
Heymsfield and colleagues measured the energy content of each kilogram actually lost during voluntary caloric restriction and found it rises steeply over the opening weeks: roughly 858 ± 388 kcal/kg at week 4, climbing toward about 6,000 kcal/kg by week 6, with differences between men and women (Metabolism, 2012). The early kilograms are cheap because they are largely water and glycogen, not fat. That is the first-week drop nobody can repeat in week five.
So the rule is wrong at both ends and in opposite directions: too slow at the start, too fast from then on, and increasingly so. The overall error is not random noise around a good average. It is a systematic drift in one direction that compounds the longer you extrapolate.
What to use instead
There is a defensible replacement, and it is free. The US National Institutes of Health publishes the Body Weight Planner through NIDDK, built on Hall's dynamic model. It is the institutional answer to static-deficit arithmetic, and unlike the 3,500 rule it predicts the plateau instead of pretending it will not arrive.
Two practical rules follow from all of the above:
- Use the 3,500 rule for weeks, not years. Over a few weeks it is close enough to sanity-check a plan. Every additional month you extrapolate makes it more wrong, in a predictable direction.
- Do not model the curve — re-measure the point. If your expenditure falls as you get lighter, the fix that requires no modelling at all is to recompute your target once your weight has actually changed. A fresh calculation on today's body beats a stale projection from three months ago.
That second rule is also how to set the target in the first place. If you have not, our guide to how many calories to eat per day walks through Mifflin-St Jeor and an activity factor, which is the same snapshot equation this drift eventually invalidates — hence the recomputing.
What PlateLens does about this, and what it doesn't
Being precise here matters more than being flattering. PlateLens does not model metabolic adaptation. Its targets come from Mifflin-St Jeor with an activity factor, which is a snapshot equation and drifts for exactly the reason described above. There is no algorithm inside it inferring your metabolism from your weight trend, and this article would be worthless if it claimed otherwise while explaining why such claims are hard.
What the app does is the simpler thing this article argues for: it prompts you to recompute your target when your weight has moved. That is the practical answer to a deficit that quietly shrinks underneath you — not a better projection, just a fresher starting point. If you want the number to stay current rather than modelled, that is the narrow, honest thing PlateLens is good at, alongside logging a meal from a photo, a typed or spoken description, or a barcode.
The same figure shows up in Balance, the app's energy view, which turns your average net balance into an estimated pace using this same 7,700 kcal/kg conversion. Our write-up of how Balance is calculated labels that a rule of thumb rather than a law, and points at the NIH Body Weight Planner for long-range planning. A number this shaky is an estimate wherever you meet it. PlateLens is on iOS and Android, syncs with Apple Health and Health Connect, and has a free plan with daily limits.
A note on your health. This is general information about where a widely repeated number comes from, not medical or dietetic advice. Calorie needs and safe rates of weight change differ from person to person. Talk to a doctor or registered dietitian before changing your diet or setting a deficit — particularly if you are pregnant or breastfeeding, under 18, managing a medical condition or a medication that affects weight, or have any history of disordered eating. If tracking numbers is making you anxious or you find yourself pushing your target steadily downward, that is a reason to stop and talk to someone rather than to recalculate.
Keep the number current instead of extrapolating it
PlateLens logs meals from a photo, a description or a barcode, and prompts you to recompute your calorie target once your weight has actually moved. Free plan with daily limits, on iOS and Android.
Frequently asked questions
How many calories are in a pound of fat?
The conventional figure is 3,500 calories per pound, and 7,700 per kilogram. It comes from a 1958 paper by Max Wishnofsky, who calculated the energy equivalent of a pound of body weight gained or lost assuming adequate protein intake. It is an assumption about tissue composition rather than a measured constant, and Wishnofsky himself noted the value is much less during fasting or refeeding after a fast.
How many calories are in a kilo of fat?
About 7,700 calories, the metric form of the same rule, given as 32.2 MJ/kg in Kevin Hall's 2008 modelling paper. The same caveat applies: it is a composition assumption, and Hall's modelling shows it only approximately matches people whose initial body fat is above roughly 30 kg, while overestimating the deficit required for leaner people.
Why isn't a pound of fat 4,080 calories?
Because 9 calories per gram is a food-energy conversion factor, not a property of body tissue. The FAO lists it as the Atwater general factor for metabolisable energy from dietary fat (37 kJ/g), alongside 4 for protein and carbohydrate and 7 for alcohol. A pound of pure lipid would hold about 4,080 kcal, but adipose tissue also contains water and protein, and the weight you actually lose is a mixture of fat mass and fat-free mass.
If I eat 500 calories less a day, will I lose a pound a week?
For a while, roughly. That is the arithmetic behind the 1998 NHLBI expert panel's advice of a 500 to 1,000 kcal/day deficit for about 1 to 2 lb a week. But the rule assumes your expenditure stays where it started, and it does not, because a lighter body costs less to run. Thomas and colleagues compared the rule against actual loss in seven supervised or objectively measured experiments and concluded it grossly overestimates weight loss.
Why did I lose four pounds the first week and then almost nothing?
Early weight change is largely water and glycogen rather than fat. Heymsfield and colleagues measured the energy content of each kilogram lost rising steeply over the first weeks of voluntary restriction, from roughly 858 ± 388 kcal/kg at week 4 toward about 6,000 kcal/kg by week 6, with differences between men and women. The first pounds are cheap; the later ones are not.
What should I use instead of the 3,500-calorie rule?
For anything longer than a few weeks, a dynamic model. The NIH Body Weight Planner, published by NIDDK and built on Kevin Hall's model, accounts for the plateau the static rule cannot predict. Its published rule of thumb is roughly 24 kcal/day per kilogram (about 10 per pound) of eventual weight change, with about half of that change reached in roughly a year and 95% in about three. The practical alternative is to recompute your calorie target whenever your weight has moved.