Guide

Why calorie counts are never exact

Calorie calculators and labels give figures with decimal points, and that appearance of precision is misleading. In reality both the calories taken in and those spent work with estimates, and this is the nature of the system rather than a fault in it.

Short answer

Calorie counts are estimates layered on estimates: label figures are averages and include a tolerance, energy coefficients are broad averages, the energy absorbed in digestion varies from person to person, and portion estimates can be wrong. The formulas on the expenditure side rest on population averages too. Calculations are therefore not exact measurements but a starting point to be corrected by observation.

The number on the label is an average

The “457 kcal” on a packet of biscuits is not a result measured for every packet in that batch; it is an average calculated from the typical composition of the product. Regulation allows a certain tolerance between the label figure and the real content; the allowance varies by component but can reach roughly 20 per cent. In natural products the difference is greater still: the sugar content of two apples varies considerably with variety, ripeness and size.

The energy coefficients are averages too

The values of 4 kcal per gram for protein and carbohydrate and 9 kcal per gram for fat come from the Atwater system, developed roughly a century ago. These coefficients are broad averages, and real values deviate somewhat between different foods within the same macronutrient group. The system is good enough for everyday use, but it is worth knowing that the digits after the decimal point carry no meaning. What the coefficients mean is explained in the calorie guide.

Not every calorie you eat is absorbed

The energy on a label describes the total potential inside the food; how much of it the body actually takes up is a separate matter. Absorption can be lower in foods high in fibre and in foods whose cell walls stay intact. There are studies showing that the measured usable energy of whole almonds is noticeably lower than the label figure, for example. Cooking, grinding and chewing increase absorption. Digestive systems and gut flora also work differently from person to person. The same plate may therefore not deliver exactly the same net energy to two people.

Portion estimates: the biggest source of error

Research shows that people usually underestimate how much they have eaten; portions judged by eye, oil poured freely and tastings along the way do not enter the account. A tablespoon of olive oil judged by eye easily becomes a tablespoon and a half. Confusion between raw and cooked weight can also throw an estimate out by two or three times; that trap is described in detail in the raw and cooked weight guide. On the expenditure side, BMR and TDEE formulas rest on averages, and the calorie estimates of watches and bands are only rough indicators.

When the uncertainties stack up

Taken one at a time, each margin of error looks tolerable; the problem is that they stack up in the same calculation. Take a simple example: a product whose label figure is genuinely 10 per cent out, eaten in a portion you have estimated 20 per cent smaller than it really was, with part of its energy passing through unabsorbed. When the three layers work in the same direction, the gap between the number in your notebook and the energy entering your body can comfortably approach a third. Fortunately errors do not always accumulate in the same direction; you overestimate one item and underestimate another, and part of the difference cancels out in the daily total. That statistical evening-out is why calorie tracking continues to be useful: the number for a single meal is unreliable, while the average of records kept over weeks shows a meaningful trend. The lesson is this: trust the pattern of the numbers rather than any one of them, and base your decisions on weeks rather than days.

How to work with this uncertainty

Uncertainty does not mean the calculations are useless; what is wrong is the way they are used. The workable approach is this: treat the figures as estimated ranges, look at how things go over a few weeks rather than on a single day, and update your decisions in small steps according to what you observe. Following the numbers as though they were exact facts is both misleading and tiring, and if number-tracking causes you anxiety, agreeing a method that does not require numbers with a dietitian is a sounder route. Where there is a chronic illness, a pregnancy or a history of an eating disorder, decisions about energy intake should always be made with a specialist.

Accepting the uncertainty is in fact a liberating step: you are freed from the burden of decimal points and can move your attention to the areas you genuinely control, such as portion awareness and variety in what you eat.

Frequently asked questions

If it is not exact, is calorie tracking pointless?

No. Even as an estimate, tracking builds awareness of portions and eating patterns. What is pointless is treating the numbers as exact to the gram and reading large meanings into small differences.

Which source of error is the largest?

In everyday life the biggest deviation usually comes from portion estimates. Label tolerances and differences in coefficients are relatively small, while portions judged by eye can produce errors of over fifty per cent.

Can I trust the calories my smartwatch says I have burned?

The exercise calorie estimates of such devices are rough indicators and can deviate considerably from person to person. They are useful for comparing workouts, not for adding directly into a calorie calculation.

Why do two apps give different figures for the same dish?

Apps use different databases and different recipe assumptions. The amount of oil, the portion size and the cooking method of a dish with the same name vary from entry to entry, and the difference comes from there.

This guide draws on the known limits of the Atwater energy coefficients, on the general framework of food labelling tolerance rules and on the general findings of published research into nutrient absorption.

First published: 2026-08-12Last reviewed: 2026-08-12Editorial status: working editionReport an error