Net Carbs Calculator

The right way to compute net carbs

The "subtract fiber from total carbs" rule is half the story. Modern food labels can include sugar alcohols, allulose, and resistant starches — each with its own glycemic behaviour. Treating them all identically gives you either an inflated or deflated net-carb number, and either way your post-meal glucose response won't match the prediction. The evidence-based formula:

Net carbs = Total carbs − Fiber − Erythritol − Allulose − (½ × other sugar alcohols)

Per-ingredient evidence

Fiber (subtract 100%): Human digestive enzymes cannot hydrolyse the β-1,4 glycosidic bonds in cellulose or the soluble fibers (pectin, inulin, β-glucan, psyllium). The colonic fermentation yield is ~2 kcal/g (FDA 2016 update) but produces short-chain fatty acids, not glucose. Blood glucose impact: essentially zero.

Erythritol (subtract 100%): ~90% absorbed in small intestine and excreted unchanged in urine within 24 hours. Multiple CGM and OGTT studies (Bornet 1996, Munro 1998, and more recent CGM trials) show no measurable glucose or insulin response to doses up to 50g.

Allulose (subtract 100%): A rare sugar (D-psicose), structurally similar to fructose but ~70% absorbed and ~75% excreted unchanged. The FDA exempted allulose from the "Added Sugars" and "Total Sugars" lines in 2019 specifically because of its negligible glycemic response. Some controlled studies (Iida 2008) actually show slight glucose-lowering effects.

Maltitol, xylitol, sorbitol, isomalt, lactitol (subtract 50%): Published glycemic indices range from 7 (xylitol) to 52 (maltitol). All are partially absorbed and partially fermented. The 50% convention is endorsed by the ADA and reflects average glucose response across this class. For maltitol specifically, the response can be closer to 60–70% of an equivalent sucrose dose — if you're CGM-tracking and see a spike from a "low-carb" product loaded with maltitol, that's why.

Resistant starch (subtract 100% if listed): Cooled cooked starch (like rice or pasta cooled overnight) develops resistant starch, which behaves like fiber. The FDA allows manufacturers to list it on labels separately or as part of fiber.

Why label "net carbs" claims are often wrong

Commercial keto bars often advertise "2g net carbs" by subtracting 100% of their sugar alcohols, regardless of which alcohol it is. A bar with 30g total carbs, 8g fiber, and 20g maltitol gets marketed as 2g net carbs. Using the partial-subtraction approach: 30 − 8 − (½ × 20) = 12g. CGM users typically see post-meal glucose responses that match the 12g estimate, not the 2g label claim. This calculator gives you the realistic number.

Net carbs for insulin dosing in type 1 diabetes

Most CDEs and endocrinologists teach carb counting using net carbs for insulin-to-carb ratios. The 2019 ADA Standards of Care explicitly endorse this for fiber-rich meals: foods with ≥5g fiber per serving should be counted using net carbs, not total carbs, for bolus calculation. For sugar-alcohol-containing foods, the partial subtraction prevents undercounting and the hypoglycemia that follows. Always discuss with your care team before changing carb-counting methodology.

Related tools

• Glycemic Load Calculator
• A1C ↔ eAG Converter
• Macro Calculator
• The 4-4-9 Rule — Where It Breaks Down

Frequently asked questions

Are net carbs an official nutrition label term?

In the US, "net carbs" is not regulated by the FDA. The label requires Total Carbohydrate, Dietary Fiber, Total Sugars, Added Sugars, and (when present) Sugar Alcohols. Manufacturers compute "net carbs" themselves and the methodology varies — some subtract all fiber and all sugar alcohols (inflated subtraction), some only subtract fiber, some only subtract erythritol. Canada and the EU use slightly different conventions. The calculation here uses the evidence-based partial subtraction approach: 100% of fiber, 100% of erythritol, 50% of other sugar alcohols.

Why doesn't erythritol count as a carb when it's 4 calories per gram on the label?

US labels are required to count erythritol toward Total Carbohydrate by weight, but unlike sugar it doesn't raise blood glucose meaningfully — the FDA acknowledged this in a 2020 guidance. Approximately 90% of ingested erythritol is absorbed in the small intestine and excreted unchanged in urine within 24 hours; the small remaining amount is fermented in the colon with minimal energy yield. Multiple controlled feeding studies (Munro 1998, Bornet 1996, and more recent CGM studies) show essentially zero post-prandial glucose response to erythritol doses up to 50g. Caveat: the 2023 Witkowski et al. Nature Medicine paper raised concerns about cardiovascular signalling from very high circulating erythritol — relevance to dietary intake is still debated.

Why subtract only 50% of maltitol?

Maltitol has a real glycemic response — published GI estimates range from 35–52 (compared to glucose 100, sucrose 65). Per gram it provides roughly 2.1 kcal (vs sucrose 4 kcal) but it does raise blood glucose and insulin in proportion to its partial absorption. The 50% subtraction convention used by the American Diabetes Association and most keto-aware diabetes educators reflects this: half "counts" as a real carb. Xylitol and sorbitol have slightly lower glycemic responses (GI 7–13 and 9, respectively) but similar partial absorption — 50% is a reasonable practical default for all non-erythritol polyols.

Do I subtract all fiber or only insoluble fiber?

All fiber. Both insoluble fiber (cellulose, wheat bran) and the bulk of soluble fiber (psyllium, oat beta-glucan, inulin, FOS) pass through the small intestine undigested by human enzymes. Some soluble fiber is fermented by colonic bacteria to short-chain fatty acids, which the body can use for energy (~2 kcal/g) — but it does not raise blood glucose. The 2016 FDA decision to count fiber at 2 kcal/g (down from the 4 kcal/g Atwater default) reflects this. Total carbs − total fiber is correct for net-carb purposes.

Net carbs vs total carbs — which should I track for diabetes?

For carb counting at meals (the insulin-dosing calculation in T1D and intensive insulin T2D): net carbs are the better predictor of post-meal glucose response. The ADA explicitly endorses this for insulin dosing. For general carb-restricted diets without insulin: either works as long as you're consistent. CGM users will see that net carbs predict the glucose peak much better than total carbs, especially in foods that pretend to be low-glycemic (sugar-alcohol-laden keto bars often spike more than the "net carbs" claim suggests because manufacturers over-subtract).