Net Carbs: What the Subtraction Assumes

Three terms, three decisions

The formula looks like accounting: take the carbohydrate line, take away the fibre, take away the sugar alcohols, and what is left is what your blood sugar will see. Read it again and notice that not one of the three quantities is a direct measurement of anything. Total carbohydrate is a residual in the United States and a category definition in Europe. Fibre is a class of molecules defined by what does not happen to them in the small intestine, and the analytical method you use decides which molecules qualify. Sugar alcohols are a family whose members behave nothing like each other. The subtraction is a model of digestion, dressed as arithmetic.

That does not make it useless. A model can be crude and still be better than nothing, and for someone deciding between two chocolate bars, net carbs is a rougher instrument than a glucose meter but a sharper one than the total carbohydrate line alone. The point is to know which assumptions you have quietly signed up to, because two of them fail loudly — one at the border between labelling regimes, and one inside the sugar-alcohol term.

The United States and the European Union do not mean the same thing by "carbohydrate"

The American definition is a subtraction of its own. 21 CFR 101.9(c)(6) says total carbohydrate content is calculated by subtracting the sum of crude protein, total fat, moisture and ash from the total weight of the food. Nothing about carbohydrate is measured; it is whatever is left over. Because dietary fibre is not protein, fat, water or ash, it survives that subtraction and sits inside the total — which is precisely why the US Nutrition Facts panel then declares Dietary Fiber as an indented line underneath Total Carbohydrate.

The European definition is a positive one. Annex I of Regulation (EU) No 1169/2011 defines carbohydrate as any carbohydrate which is metabolised by humans, and specifies that it includes polyols. Fibre gets its own definition in the same annex — carbohydrate polymers of three or more monomeric units that are neither digested nor absorbed in the human small intestine — and Annex XV puts it on its own line in the nutrition declaration, below carbohydrate rather than inside it. The consequence is not subtle: on an EU label the number printed against "carbohydrate" has already excluded the fibre, so the reflex to subtract fibre from it removes the same grams twice. Do that with almonds and you get −3.4 g, which is the arithmetic telling you the model has been applied to the wrong input.

One food, two labels: almonds

USDA FoodData Central gives raw almonds, per 100 g, as 21.2 g protein, 49.9 g fat, 4.41 g water and 2.97 g ash. Apply the regulation: 100 − (21.2 + 49.9 + 4.41 + 2.97) = 21.5 g, and the database prints 21.6 g of total carbohydrate, the difference being rounding in the component values. Of that, 12.5 g is dietary fibre and 4.35 g is sugars. A US label therefore reads Total Carbohydrate 21.6 g, Dietary Fiber 12.5 g, Total Sugars 4.4 g — or, on a 1-ounce serving, 6.1 g, 3.5 g and 1.2 g.

Now write the European declaration for the same nuts. Carbohydrate means the metabolised fraction, so the figure is 21.6 − 12.5 = 9.1 g, of which sugars 4.4 g, with fibre declared separately at 12.5 g. Both labels describe an identical handful of almonds and both are correct in their jurisdiction. The low-carb reader wants 9.1 g per 100 g in either case — but on the US label they get there by subtracting, and on the EU label they get there by reading. Subtracting on the EU label is the single most common way the calculation goes wrong, and it goes wrong in the direction that flatters the food.

Sugar alcohols are not one ingredient

Erythritol and maltitol sit on the same label line and behave nothing alike. EFSA's re-evaluation of erythritol reports a high degree of absorption from the small intestine — studies putting it in the range of 60% to 90% — after which it is excreted unchanged in the urine, and the panel found that the available evidence consistently showed no short-term effect on postprandial glucose homeostasis. Maltitol is a different story: Livesey's 2003 review in Nutrition Research Reviews tabulates a glycaemic index of 35 for maltitol against 0 for erythritol, on a scale where glucose is 100. Subtracting both at 100% treats a substance with roughly a third of glucose's glycaemic index as if it had none.

European law has already priced the difference into energy. Annex XIV of Regulation 1169/2011 assigns 4 kcal/g to carbohydrate except polyols, 2.4 kcal/g to polyols, 2 kcal/g to fibre — and a separate line giving erythritol 0 kcal/g. That is a regulator saying, in the only language a label speaks, that these are not the same substance. A subtraction that flattens them into one term is discarding a distinction the law itself makes.

A chocolate bar worked both ways

Take a no-added-sugar dark chocolate with a composition typical of the category: per 100 g, 38 g fat, 8 g protein, 1 g water and 2 g ash. By the US rule that gives 100 − 49 = 51 g of total carbohydrate, made up of 12 g fibre, 33 g maltitol, 4 g sugars and 2 g of starch and minor carbohydrate. The US label reads Total Carbohydrate 51 g, Dietary Fiber 12 g, Total Sugars 4 g, Sugar Alcohol 33 g, and the front of pack advertises 51 − 12 − 33 = 6 g of net carbs. The EU label for the identical bar reads Carbohydrate 39 g, of which sugars 4 g, of which polyols 33 g, and Fibre 12 g. Same product, two true declarations, and only one of them is compatible with the subtraction people have memorised: run it on the EU label and you get 39 − 12 − 33 = −6 g.

Now change one thing. Swap the 33 g of maltitol for 33 g of erythritol and leave everything else alone. The advertised net carbs figure does not move — still 6 g — because the subtraction cannot see the difference. The declared energy does move: applying the Annex XIV factors gives 501 kcal per 100 g with maltitol and 422 kcal with erythritol, a gap of 79 kcal that comes entirely from 33 g at 2.4 kcal/g against 33 g at zero. And on Livesey's figures, the maltitol version carries a glycaemic load of about 12 from the polyol alone, against zero for the erythritol version, on top of the 2.6 contributed by the 4 g of sugars. The number on the front of the packet is identical for two products that differ by 79 kcal and by essentially all of their glycaemic impact.

No regulator defines "net carbs" — anywhere

Total carbohydrate, dietary fibre, sugars and sugar alcohols are all defined terms with mandated calculation rules. "Net carbs" is not among them. It appears in no part of 21 CFR 101, it appears in no annex of Regulation 1169/2011, and it therefore has no compulsory method behind it. A manufacturer choosing which sugar alcohols to subtract, and at what rate, is making an editorial decision about their own packaging, not complying with a standard.

That is why the term does double duty as a marketing device. Its whole appeal is that it produces a smaller number than the mandatory one, and the discretion in how it is computed all points the same way. There is no version of the calculation in which a manufacturer discovers, to their surprise, that their bar has more net carbs than they thought. Treat any front-of-pack net-carb claim as the lowest defensible number the maker could construct, and recompute it yourself from the regulated lines on the back, which are the only figures anyone had to stand behind.

What the subtraction is legitimately good for

It is a comparator, not a measurement. Two products in the same category, on the same labelling regime, with the same kind of sugar alcohol: ranking them by net carbs is more informative than ranking them by total carbohydrate, because it strips out a component that genuinely does not reach the bloodstream as glucose. That is a real and modest job, and the subtraction does it competently.

What it is not for is dosing insulin. Carbohydrate counting for diabetes is a clinical skill taught with its own conventions, and different teams handle fibre and polyols differently for good reasons that depend on the individual, the insulin and the meal. If you carry that decision, it belongs with your diabetes team and your own post-meal readings, not with a subtraction copied off a packet — and any change to how you count should be made with them, never on the strength of an article.

Frequently asked questions

Why does my net carb figure come out negative on a European label?

Because you subtracted the fibre twice. Under Regulation 1169/2011 the carbohydrate figure already excludes fibre, which is declared on its own line. On an EU label the carbohydrate number is the net-of-fibre number; subtract only the polyols, if you subtract anything at all.

Should I subtract all the sugar alcohols, or only some of them?

The honest answer is that they are not interchangeable. Erythritol is largely absorbed and excreted unchanged, with no measurable short-term effect on postprandial glucose in EFSA's assessment, so subtracting it in full is defensible. Maltitol has a published glycaemic index of 35 against glucose's 100, so subtracting all of it is not. If the ingredient list does not name which polyol is used, treat the net figure as an upper-bound claim.

Is "net carbs" a legally defined term?

No. Total carbohydrate, dietary fibre, sugars and sugar alcohols all have mandated definitions and calculation rules; net carbs has none, in the United States or the European Union. It is a computed figure whose method the manufacturer chooses, which is why it is best recomputed from the regulated lines rather than trusted from the front of the pack.

Why do the same almonds show 21.6 g of carbohydrate in one database and about 9 g in another?

Because the two databases follow different labelling conventions. A US-style figure is carbohydrate by difference and contains the 12.5 g of fibre; a European-style figure counts only metabolised carbohydrate and puts the fibre on its own line. Neither is wrong. Before comparing two sources, check whether their carbohydrate column includes fibre.

Two bars both claim 6 g of net carbs. Are they equivalent?

Not necessarily. In the worked example above, swapping 33 g of maltitol for 33 g of erythritol leaves the claimed 6 g untouched while moving the declared energy from 501 to 422 kcal per 100 g and removing essentially all of the polyol's glycaemic load. Read the ingredient list and the energy line, which are regulated, before trusting the claim, which is not.

Can I use net carbs to decide my insulin dose?

That is a clinical decision, not a labelling one. Carbohydrate counting for diabetes is taught with its own conventions, and teams differ in how they handle fibre and polyols for reasons specific to the person, the insulin and the meal. Never change how you count on the strength of a web page — take it to your diabetes team, who can see your own post-meal readings.