Gluten-free labeling

21 CFR 101.91 defines "gluten-free" as a food that (a) inherently does not contain gluten, or (b) does not contain an ingredient that is a gluten-containing grain (wheat, rye, barley, or a crossbred hybrid like triticale), or (c) does not contain an ingredient derived from a gluten-containing grain that has not been processed to remove gluten, or (d) does contain such an ingredient but the final food contains less than 20 ppm gluten. The single quantitative threshold is < 20 ppm. The terms "no gluten," "free of gluten," and "without gluten" carry the same regulatory meaning as "gluten-free."

The 20 ppm threshold derives from the lowest level at which validated analytical methods can reliably quantify gluten and from clinical evidence on celiac-disease tolerance. FDA's preamble to the 2013 rule explained that 20 ppm is below the level at which most individuals with celiac disease experience adverse effects, while being analytically achievable. Codex STAN 118-1979 sets the same threshold internationally. Third-party certification programs typically set a stricter standard: GFCO (Gluten-Free Certification Organization) certifies to 10 ppm; the Celiac Support Association ASCC certifies to 5 ppm.

The FDA-recognised analytical method is the R5 ELISA Mendez sandwich assay (AOAC 2012.01 / AACC 38-50.01), which uses a monoclonal antibody to the QQPFP epitope of prolamins (gliadin in wheat, secalin in rye, hordein in barley). The R5 sandwich format quantifies intact prolamins; the R5 competitive format quantifies fragments and is required for fermented / hydrolyzed foods (per the 2020 rule). Detection range typically 5-80 ppm; LOQ typically 5 ppm.

• Sample preparation must homogenize; gluten distribution in finished foods is often heterogeneous and a single point sample can mislead.
• Sampling plan must reflect risk: composite sampling for low-risk products; multiple discrete samples for high-risk (high cross-contact, low-volume) products.
• Method validation per AOAC / AACC for the specific matrix; matrix interferences (high-fat, high-starch, high-protein matrices) can suppress or enhance signal.
• Fermented / hydrolyzed foods (soy sauce, beer, vinegar, hydrolyzed protein) cannot be tested by R5 sandwich format — fragments below the sandwich threshold but still antigenic; use R5 competitive ELISA + records demonstrating the source materials were < 20 ppm before fermentation / hydrolysis.
• Distilled products (vodka, whiskey, distilled vinegar) are presumed gluten-free regardless of source by FDA — distillation removes prolamins.

FALCPA wheat-allergen labeling (21 U.S.C. §343(w)) and gluten-free labeling (21 CFR 101.91) are independent obligations:

Oats are not inherently gluten-containing — avenin (the oat prolamin) is structurally distinct from wheat / rye / barley prolamins and is tolerated by most (not all) people with celiac disease. However, conventional oats are typically rotation-cropped, harvested, transported, and processed on shared equipment with wheat / barley, resulting in cross-contact contamination commonly at 200-2000 ppm gluten. Conventional oats therefore cannot bear a gluten-free claim. "Gluten-free oats" must be sourced through a validated supply chain — typically purity-protocol (PP) production with dedicated land, harvest, storage, and milling; or sorted gluten-free (mechanical / optical / colour sorting) with downstream validation. Either approach requires supplier-side qualification, finished-product testing, and ongoing surveillance.

• Claim on label, no analytical validation — most-common Warning Letter pattern in the gluten-free space. The 20 ppm threshold is quantitative; if you cannot test, you cannot claim.
• Wheat flour in product + gluten-free claim — pathway 3 not available for wheat flour; immediate misbranding.
• Conventional (uncertified) oats + gluten-free claim — cross-contact risk virtually guarantees > 20 ppm.
• Cross-contact at manufacturing — same equipment runs gluten-containing and gluten-free products without validated cleaning + testing. Allergen-control program (§117.135) must validate that cleaning + flushing brings carryover < 20 ppm.
• Single composite sample as the validation evidence — heterogeneity in finished products can yield individual units > 20 ppm even when composite is < 20 ppm. Sampling plan must reflect risk.
• Fermented / hydrolyzed product validated by R5 sandwich (not competitive) — sandwich format misses fragments; result reads as < 20 ppm but actual antigenic load is much higher.
• Source ingredient gluten-free but supplier doesn't certify — finished-food testing each batch is the only defence.
• Distillation assumption applied to non-distilled product — only distilled products carry the FDA distillation presumption; filtered, ultra-filtered, or otherwise processed do not.
• GFCO / certified-claim packaging copy on non-certified product — using a third-party certification mark without actual certification is trademark infringement on top of misbranding.
• Recipe / supplier change without re-validation — single ingredient swap can change the gluten profile; change control must trigger re-test.

• Labeled-claim register: every product carrying "gluten-free" / "no gluten" / "free of gluten" / "without gluten" registered with chosen qualification pathway (inherent / no-gluten-ingredient / processed-derivative / finished-food testing).
• Per-lot analytical evidence: R5 ELISA result per batch (or per defined sampling frequency); LOD / LOQ / matrix-validated; method version controlled; result < 20 ppm gate before release.
• Sampling plan: per-product sampling plan defined (frequency, n, location); composite vs discrete; for high-risk matrices forced discrete sampling.
• Fermented / hydrolyzed flag: products in this category forced to R5 competitive ELISA + source-material < 20 ppm evidence.
• Allergen-control program linkage: gluten-free products linked to allergen-control matrix; same-equipment-as-wheat-rye-barley flagged; validated cleaning protocol required; post-cleaning swab + flush testing < 20 ppm gate.
• Supplier-certification tracker: ingredients sourced as gluten-free (gluten-free oats, wheat starch, certified flours) tracked with supplier CoA + supplier certification + audit-date; expired certifications hard-block consumption.
• FALCPA wheat statement: dual-label validation — wheat-containing products forced to carry "Contains wheat" regardless of gluten-free status.
• Recipe / supplier change control: any change to ingredient, supplier, recipe, or equipment in a gluten-free product opens a change-control workflow that forces re-validation before the next claim-bearing batch.
• Certification-mark control: if product carries GFCO / NSF Certified / ASCC mark, certification number + expiry tracked; expired certifications hard-block use of the mark.
• Class-action exposure register: every gluten-free claim with weak evidence (composite-only sampling, sole-source supplier certification, fermented / hydrolyzed with sandwich-only data) surfaced for legal review.
• Periodic verification: quarterly summary of all gluten-free products, sampling plan adherence, exceedance trend, and corrective actions for senior officer review.