Four numbers no new diet will ever change:
These are the Atwater coefficients, the operational values used by USDA, EFSA, and FAO/WHO. They sit behind every label in the store and every calorie calculator. They've held since 1899 because they rest on combustion physics (bomb calorimeter) plus averaged digestibility in the small intestine. Not on a trend, not on a nutritionist's opinion.
In this article we explain where they came from, why they're stable, and how to use them for calorie counting. For a deeper dive on each macro we have separate pages, linked at the end.
| Macronutrient | kcal/g (Atwater) | kcal/g (calorimeter) | Digestibility | TEF (thermic effect) |
|---|---|---|---|---|
| Protein | 4 | 5.65 | 80–95% | 20–30% |
| Fat | 9 | 9.4 (animal) / 8.3 (MCT) | 95–98% | ~3% |
| Carbs | 4 | 3.79–4.16 | 95–98% (sugars, cooked starch) | 5–10% |
| Alcohol | 7 | 7.07 | ~100% | 15–20% |
| Soluble fiber | ~2 | 4.0 (if digested) | ~50% (gut fermentation) | n/a |
| Insoluble fiber | ~0 | 4.0 | ~0% | n/a |
Step 1: bomb calorimeter. A food sample is burned in oxygen inside a sealed chamber. The released heat warms surrounding water. By measuring water temperature increase, we know how many kcal the sample released. That gives gross values:
Step 2: digestibility and metabolic losses. Your body doesn't burn the sample 100%. Part passes through the gut unabsorbed, part is excreted in urine. Atwater measured average losses at the turn of the 20th century:
Hence 4 / 9 / 4 / 7. These are operational values for diet: they account for actual utilization, not just physical energy content.
Three reasons:
Any "revolutionary diet" claiming "calories don't count" or "1 g of carbs is 0 kcal for some people" is lying. Counting to the nearest 1 kcal isn't worth it because biological noise is bigger. But the 4/9/4/7 coefficients are harder than 99% of what you'll read about nutrition.
EU product labels must show:
Energy: 380 kcal / 100g
Fat: 8 g
of which saturates: 3 g
Carbohydrates: 50 g
of which sugars: 12 g
Protein: 24 g
Salt: 1.2 g
Sum check:
8 × 9 + 50 × 4 + 24 × 4 = 72 + 200 + 96 = 368 kcal
The label shows 380, and the 12 kcal difference can come from fiber (~3 g in 50 g of carbs, ~6 kcal) and sugar alcohols. Generally: if you check (F × 9) + (C × 4) + (P × 4) and get a value close to the listed energy, the label is consistent.
BiteNote does this verification automatically on every log if you send a label photo or a product name from OpenFoodFacts.
📸 → BiteNote: ham and cheese sandwich, ~340 kcal (320–360). Protein 18g (72 kcal), Fat 14g (126 kcal), Carbs 36g (144 kcal). Confidence: medium.
Send a photo, a voice note, or a line of text, and the answer comes straight back in the chat. Nothing to install: you already have Telegram, and /today shows the whole day broken down by macro without leaving the conversation.
From two steps: a bomb calorimeter measures combustion energy (~5.65/9.4/4.0/7.07 kcal/g gross), then Atwater subtracted metabolic losses and indigestible fractions, giving net values 4/9/4/7. Standardized by FAO/WHO in 1947 and 2003 (Report 77).
Within precision (~5%), no. Athletes have slightly better protein digestibility (up to 95–97%), older adults slightly worse (~85% with pancreatic dysfunction). For everyday calorie counting: 4/9/4/7 applies. Individual calibration only matters in clinical diets with a registered dietitian.
Soluble fiber (inulin, beta-glucan, pectins) sits at ~2 kcal/g, partly fermented by gut bacteria. Insoluble fiber (cellulose) is ~0 kcal/g. Sugar alcohols (polyols): erythritol ~0.2, xylitol ~2.4, maltitol ~2.1 kcal/g. EU labels must distinguish them, and BiteNote includes them correctly in total kcal.
It can show slightly different values, because manufacturers have a ±20% tolerance in the EU. A pierogi labelled "30 g" is really 26–35 g, so the label is already an approximation of the food in front of you.
The argument is insulin-based: since carbs raise insulin, and insulin blocks lipolysis (fat breakdown), cutting carbs lowers insulin and mobilizes fat. Modern consensus (e.g., Hall et al. 2023 meta-analysis): the insulin effect exists but is smaller than keto advocates claim. Daily energy balance (kcal in vs kcal out) remains the main lever. The 4/9/4/7 coefficients don't change on keto.
They can say you have different digestibility (pancreatic disorders, IBS, microbiota) and that you should adjust. But the 4/9/4/7 coefficients as the calculation base remain fixed: adjustments are at the ±5–10% level, not ±50%. If someone says otherwise, they're probably selling supplements.
Log meals in 3 seconds on Telegram → @bitenotebot