How Net Carbs Are Actually Calculated (and Where Labels Lie)

The Testing Protocol post gave the formula in one line: total carbohydrates minus fiber minus half of any sugar alcohols. That’s the number every scorecard in this lab runs on, so it’s worth being precise about where that formula comes from, why “half” and not “all,” and — the part that actually matters when you’re standing in an aisle reading a keto snack’s label — where the number on the package can undercount the truth even when nobody’s technically lying.

The formula, and why it’s shaped the way it is

“Net carbs” isn’t measuring a different nutrient than “total carbs.” It’s a claim about which carbs your body actually processes for energy, built on two exclusions:

• Fiber gets subtracted because the human gut doesn’t have the enzymes to break most dietary fiber down into glucose. It passes through largely intact (with some fermented by gut bacteria into short-chain fatty acids, a separate metabolic story). Total carbohydrate on a label already includes fiber — it’s counted, then subtracted back out.
• Sugar alcohols get partially subtracted because they’re absorbed and metabolized incompletely and inconsistently across different molecules. Erythritol is the extreme case: it’s absorbed in the small intestine but barely metabolized at all, and the majority is excreted unchanged in urine — which is the standard justification for treating it as close to zero net-carb impact, even though it technically carries a small caloric value (roughly 0.2 kcal/g, versus 4 kcal/g for glucose). Other sugar alcohols — maltitol especially — are metabolized more like real sugar, which is exactly why lumping “sugar alcohols” into one bucket and subtracting all of it isn’t honest math. The “subtract half” convention is a rough, deliberately conservative compromise for exactly this reason: it doesn’t pretend maltitol behaves like erythritol.

It isn’t an official term — which is the whole problem

Here’s the detail that changes how you should read a keto product’s front-of-package claim: “net carbs” has no legal definition. “Total carbohydrate,” “dietary fiber,” and “total sugars” are all defined and regulated — in the US under 21 CFR 101.9, in India under FSSAI’s nutrition labelling rules. “Net carbs” is not defined in either framework. It’s an industry convention, not a regulatory category — which means there’s no single correct way to calculate it, and no requirement that a manufacturer’s on-package number be calculated any particular way. One nutrition director at Columbia University put it bluntly to a health outlet covering the issue: these terms “have been made up by food companies.”

In practice that shows up as real inconsistency: some brands subtract fiber only, some subtract fiber plus every sugar alcohol at full weight, some pick and choose which fiber fractions count. A “3g net carbs” claim from two different brands can be built on genuinely different math, applied to genuinely different ingredients, and there’s no regulator checking either one against a standard — because there isn’t one.

Where the label itself can undercount

Even before you get to the “net carbs” marketing claim, the regulated numbers feeding into it have built-in rounding gaps:

• US labels round dietary fiber to the nearest gram, and anything under 0.5g can be declared as 0g. Total carbohydrate follows the same nearest-gram rounding. That means a product with, say, 0.4g of fiber per serving is allowed to print “0g” — not wrong, per the regulation, but it means the fiber-subtraction step of your own net-carb math starts from a number that’s already been rounded away.
• Sugar alcohols aren’t always broken out at all. Under the US rule, a sugar alcohol sub-listing is only required if the manufacturer makes a claim about sugar alcohols, or about sugar/calorie content tied to them. No claim, no obligation to show the number — even though it’s still folded into the (regulated) total carbohydrate figure above it. Total carbs isn’t lying about the magnitude; you just can’t back out how much of it is sugar alcohol versus starch without the brand telling you.
• India’s FSSAI rules follow a broadly similar shape: nutrition information is required per 100g/100ml or per serving (energy, protein, carbohydrate, fat, and — per FSSAI’s own nutrition-labelling guidance — sugars called out within carbohydrate), with macronutrients rounded to one decimal place. Dietary fiber specifically only has to be declared when the product carries a fiber-related claim. We couldn’t find an FSSAI-published rounding threshold for sugar alcohols/polyols as precise as the US 0.5g rule during this fact-check — treat that gap as unresolved rather than assumed identical to the US rule.
• Serving size is the lever behind most of this. Rounding thresholds apply per serving, not per 100g. Shrink the declared serving size and more true grams fall under the “rounds to zero” line — nothing on the label is false, the math on it is just being done on a smaller slice of the product than you might assume you’re eating.

None of this requires a manufacturer to do anything illegal. It requires them to round in the direction that’s most flattering, at every point in the process where a regulation gives them room to — and regulations giving room to round is not the same as a regulation requiring honesty about the net effect.

What this means for our numbers

This is why the lab’s own net-carb figures are sourced from brand labels where a specific product is named, and USDA data for generic ingredients — not from a brand’s own “net carbs” marketing claim, which is exactly the unregulated, inconsistently-calculated number this post just spent several sections taking apart. Where we cite a manufacturer’s label figure directly, that’s the total carbohydrate and fiber lines (the regulated, defined numbers), with our own fiber/half-sugar-alcohol formula applied on top — not whatever “net carbs: Xg” is printed on the front of the bag.

It’s also why this whole site treats net carbs as directional kitchen math, not a medical carb-counting reference. Given everything above — rounding gaps, undisclosed sugar alcohol breakdowns, zero regulatory definition of the term itself — a number that claims more precision than “roughly this many grams” would be false precision dressed up as rigor. That’s the same discipline behind the “directional, not falsely precise” rule this lab already runs on.

---

Sources consulted (fact-checked 2026-09-04):

• 21 CFR § 101.9 — Nutrition labeling of food (eCFR, current) — dietary fiber and total carbohydrate rounding to the nearest gram, sub-0.5g declared as 0g; sugar alcohol sub-listing triggered only by a sugar/sweetener/calorie-related claim.
• FSSAI, Guidelines on Nutrition Labelling (CAC/GL 2-1985, adapted) — per-100g/serving nutrition declaration requirements; dietary fiber declaration required only where a fiber claim is made. Macronutrient rounding to one decimal place per current FSSAI labelling-requirement summaries; no FSSAI-specific polyol rounding threshold located during this fact-check — flagged as an open gap above rather than assumed.
• Multiple nutrition-journalism sources (Healthline, WebMD, and industry/legal-compliance coverage of FDA low-carb labeling) independently confirming “net carbs” has no FDA or FSSAI regulatory definition, and that manufacturer calculation methods vary (full vs. partial sugar-alcohol subtraction, inconsistent fiber-fraction treatment).
• Erythritol absorption/excretion mechanism (absorbed, largely unmetabolized, excreted via urine; ~0.2 kcal/g) — consistent with widely cited food-science and diabetes-nutrition sources on sugar alcohol metabolism; no single caloric or excretion-percentage figure is asserted more precisely than “the large majority,” per the site’s usual caveat on contested figures.

Caveat: we did not locate an FSSAI-published numeric rounding threshold for sugar alcohols/polyols equivalent to the US 21 CFR 101.9 rule. That gap is reported above rather than assumed to mirror the US rule. If you can point us to the figure in an FSSAI regulation or gazette notification, we will add it.

Next: what gluten actually does, structurally — and why nothing tested so far fully replaces it in one ingredient. That’s the mechanism post standing between this one and Papdi Chaat actually going into the fryer.