Nutrition & choosing · Article

How recipe nutrition is calculated, and why it is never exact

The calories on a pack come from a lab test or from the recipe. Both are accepted methods, and both describe an average, not the exact pack in your hand.

Plate Mate editorial team · Published · 4 min read

Illustration with a plate and a protein icon in Plate Mate colours
Illustration

Buy the same dish twice and one pack may hold a spoon more sauce, the other a little more meat. Both labels still show the same calories and macros, because the figure describes the planned portion, not the one in your hand. It comes either from a lab that analysed a sample or from a calculation based on the recipe. Which one, and how far a label may drift, decides how much to trust it.

Lab or recipe: both are accepted

Under EU rules, declared nutrition values are average values, based on the manufacturer's analysis of the food, on a calculation from the known or actual average values of the ingredients, or on a calculation from generally established and accepted data (European Commission guidance). That is EU law, not Thai law, but it shows that recipe calculation is a normal method, not a shortcut.

A lab measures one sample very precisely, and that sample is all it describes: FAO/INFOODS training material notes that the result holds for one combination of ingredients, in general prepared by one person, and becomes less transferable with any change in ingredients or method. Repeating the test for every dish and recipe change is also expensive. A calculation is easy to update and reflects the planned recipe rather than one batch, but it relies on average ingredient data and on assumptions about cooking.

From recipe to label

  1. List every ingredient with its weight. Including frying oil, sauce, water and salt, and only the edible part: no bones, peel or trimmings.
  2. Match each ingredient to composition data. For example "chicken breast, raw" or "rice, white, dry".
  3. Add up the nutrients. Protein, fat, carbohydrate and energy for the whole batch.
  4. Correct for cooking. A yield factor is the percentage weight change from cooking (water lost from meat, water absorbed by rice). A retention factor is the share of a nutrient left after preparation, processing, warm holding or reheating; the losses are mainly vitamins and minerals.
  5. Divide by portion. Split the batch total by the number of portions, or by the finished weight to get per 100 g.

FAO/INFOODS describes several calculation systems and recommends a "mixed method": yield applied at recipe level, nutrient retention at ingredient level.

Why the real pack differs from the label

The European Commission guidance lists the reasons, from recipe values used instead of analysis and the accuracy of analysis to variation in raw materials, processing, nutrient stability and storage. For a cooked dish, three matter most besides the portioning described above.

Ingredients vary

Fat in meat, water in vegetables and sugar in fruit shift with breed, season and cut. A database value is an average, and an average is never exactly the chicken on the chopping board.

Cooking moves water and fat

Boiling rice or legumes adds water, frying meat or baking bread removes it, and deep-frying removes water and adds fat (Bognár, 2002). How much evaporates depends on hard-to-predict factors such as the pot, the temperature and the time. The USDA publishes cooking yields for the same reason, so that a cooked food's nutrients can be calculated from the raw one (USDA ARS). Raw vs cooked weight covers this in detail.

People make slips

FAO/INFOODS lists the usual ones: forgetting the water or the frying fat, not removing inedible parts, misapplying yield or retention factors, matching an ingredient to the wrong database food, or treating a missing value as zero. Frying fat also shows why the cooking method matters; see choosing cooking oils.

How far a label may be off

That is what tolerances are for: they set how far a measured value may stray from the declared one. The European Commission guidance gives these ranges for ordinary foods (not supplements), measurement uncertainty included:

NutrientAmount per 100 gAllowed difference
Carbohydrate, sugars, protein, fibreunder 10 g± 2 g
10–40 g± 20%
over 40 g± 8 g
Fatunder 10 g± 1.5 g
10–40 g± 20%
over 40 g± 8 g
Saturated fatunder 4 g± 0.8 g
4 g or more± 20%
Sodiumunder 0.5 g± 0.15 g
0.5 g or more± 20%

A food labelled with 15 g of protein per 100 g is within tolerance if a lab finds anywhere from 12 g to 18 g. The guidance adds that declared values should sit near the average across batches, not at the edge of the range, and that for fat, sugars and salt the label should not be set at the low end when the real average is higher. These are EU control guidelines. Thailand has its own nutrition labelling rules, and this article does not describe Thai tolerance requirements.

What to do with the number

Use it to compare dishes and plan your day, not to chase decimals: a few grams between two similar meals is normal variation, and over a week single portions tend to even out. Comparing meals of different sizes is covered in per pack, per 100 g, per serving, and protein in comparing protein per portion.

Calories and macros on our packs are calculated from the recipe (ingredients and weights), not measured in a laboratory. They describe the planned recipe and portion: a useful guide for choosing a meal but, like any label value, calculated or measured, not a promise of the exact content of every pack.

Sources

  1. Guidance document on tolerances for nutrient values declared on a label (December 2012) · European Commission, DG Health and Consumers · 2026-10-23
  2. Recipe and other calculations (course slides, updated January 2021) · FAO/INFOODS (U. R. Charrondiere) · 2026-10-23
  3. Tables on weight yield of food and retention factors of food constituents for the calculation of nutrient composition of cooked foods (2002) · Bundesforschungsanstalt für Ernährung (A. Bognár) · 2026-10-23
  4. USDA Table of Cooking Yields for Meat and Poultry, Release 2 (2014) · USDA Agricultural Research Service · 2026-10-23

More in: Nutrition & choosing →