21 CFR 111.70(e): Finished Dietary Supplement Specifications

21 CFR 111.70(e) is the U.S. dietary supplement rule requiring manufacturers to establish finished-batch product specifications for identity, purity, strength, composition and limits on contamination that may adulterate the batch or lead to adulteration.

The requirement is product-specific: establish specifications for each dietary supplement manufactured. It does not mean inventing a new specification for every batch. The product specification defines the expected finished condition; batch-specific evidence supports evaluation of a particular batch against that applicable specification. This provision belongs to the dietary supplement framework in 21 CFR Part 111, not a universal finished-product standard for every regulated industry.

In practical manufacturing usage, keep four objects distinct:

• Specification: the expected condition or quality attribute of the finished product.
• Acceptance criterion: the decision boundary, such as an interval, maximum limit or required qualitative outcome.
• Method: the test or examination procedure used to produce evidence about the attribute.
• Result: the recorded observation for an identified sample and batch, expressed in the applicable units or qualitative terms.

These distinctions prevent a common recordkeeping error: treating a laboratory result as though it were the requirement itself. A practical batch production record connects the observation to the applicable criterion and method so a reviewer can understand the comparison.

The worked examples and recommended operating controls below illustrate industry practice; they are neither binding law nor official guidance. Software support is a separate evidence-management aid, not a substitute for human review or release decisions.

This map addresses U.S. dietary supplement manufacture. For conventional foods, drugs, devices, chemicals or non-U.S. destinations, check the applicable product and market rules rather than extending this provision by analogy.

Documents answer different questions

In manufacturing practice, these objects are complementary, not interchangeable:

• Specification versus certificate of analysis: the specification defines requirements; the certificate reports results for identified material and tests, not necessarily every requirement.
• Product specification versus manufacturing instruction: one defines the acceptable product; the other directs how to make it. An instruction alone does not demonstrate conformity.
• Acceptance criterion versus test method: one defines a passing result; the other defines how to obtain the measurement or observation.

An assay result cannot establish whether the specified package and label were used. 111.75(g) requires, at minimum, visual examination of finished packaging and labeling to determine this. Keep that evidence distinct in packaging and labeling controls.

In practical specification design, identity addresses what the product contains; purity addresses unwanted constituents relative to the intended material; strength addresses an amount or concentration; composition addresses intended constituents and proportions; and contamination limits address relevant adulteration hazards. Categories can overlap—for example, a degradation product can affect purity and strength—but each answers a distinct question.

Worked case 1 assumes a hypothetical vitamin C capsule labeled at 100 mg per capsule, with one capsule per serving. Assume pure ascorbic acid supplies the vitamin C; fill quantities exclude the shell. Every numerical criterion below is illustrative, not an FDA limit or a ready-to-use commercial specification.

Example specification: SPEC-VC100; revision: 01; product identifier: VC100-CAP; applicability date: 2026-01-01. Method identifiers below are hypothetical.

An appearance check or one ingredient marker cannot establish every aspect of finished-product identity. “Acceptable purity” cannot guide a defensible decision without named analytes, a calculation basis and a method. Each proposed analytical method needs product-specific suitability justification.

Average fill mass alone does not prove ingredient proportions: composition relies on formulation and manufacturing evidence, supported by component specifications. The contamination entries illustrate separate characteristics, not a universal supplement panel. The next section calculates the strength result and examines the assay window.

Continuing worked case 1, assume a label declaration of 100 mg per capsule, a nominal manufacturing target of 110 mg per capsule, a hypothetical release interval of 100–120 mg per capsule, and a measured finished-product assay of 108 mg per capsule. These numbers illustrate the distinction between label claim and release limit; they are not legal tolerances.

• Nominal overage = (110 − 100) ÷ 100 × 100 = 10%.
• Measured label-claim percentage = 108 ÷ 100 × 100 = 108%.
• The measured 108 mg per capsule meets the illustrative strength interval only.

The manufacturing target is a formulation intention, not an analytical finding. Ingredient potency, processing losses and blend or fill variability can separate the target from the finished result. Neither a formula overage nor a passing strength result, by itself, demonstrates identity, purity, composition or acceptable contamination levels.

For one limited stability illustration, assume a hypothetical 5% relative loss from the measured 108 mg. The modeled amount is 108 × 0.95 = 102.6 mg per capsule. This remains above the example’s 100 mg declaration, but the calculation does not establish shelf life or replace stability evidence. No storage condition or duration is established by that assumption.

In practice, justify the assay window using formulation capability, method performance, reporting basis, measurement variability and available stability knowledge. Consider measurement uncertainty in potency assays when defining the decision rule; do not widen limits retrospectively to accommodate a result.

Before comparing values, the reviewer should confirm whether the declaration is per capsule or per serving and record the capsule count used for any serving calculation. Retain the laboratory’s original value and units alongside converted values, conversion factors and reporting basis. Use the conversion example in the FAQ rather than treating differently expressed quantities as directly comparable.

As analytical practice, select limits around ingredient origin, extraction or processing history, dosage form, maximum daily intake and plausible contaminants. Consider metals, pesticide residues, residual solvents and microbial contamination where relevant, alongside cross-contamination controls. Assess the finished matrix’s analytical behavior: extraction recovery, interference and quantitation capability can affect whether a method supports the proposed limit.

Supplier and component results inform this assessment, but a raw-material limit should not simply become the finished-product limit. Account for dilution by other ingredients, concentration during processing, contributions from multiple components and daily product intake.

Worked case 2: converting exposure to concentration

Assume a botanical product’s maximum intake is two 500 mg capsules daily, with 500 mg representing the complete product mass tested per capsule. Daily intake is 1,000 mg = 1.0 g = 0.001 kg.

• Hypothetical internal lead-screening criterion: 0.5 micrograms/day, used solely for arithmetic—not a safe level or legal threshold.
• Concentration equivalent: 0.0005 mg ÷ 0.001 kg = 0.5 mg/kg.
• Hypothetical result: 0.7 mg/kg × 0.001 kg/day = 0.0007 mg/day = 0.7 micrograms/day.

The result exceeds the illustrative internal criterion; it does not support a conclusion that this screen was met. That evidence issue calls for review, not an inferred legal disposition.

Record each actual limit’s basis: applicable market, product, intake assumption, analyte, method and quantitation capability. For microbial monitoring, distinguish numerical counts, such as CFU/g, from specified-organism tests reported as detected or not detected in a defined test quantity; these are different decision formats.

Under 21 CFR 111.75(c), finished dietary supplement verification covers either every finished batch or a subset identified through a sound statistical sampling plan. The choice concerns which batches undergo verification—not a reduced set of product requirements.

Select tests that support the whole specification

21 CFR 111.75(c)(1) requires selecting one or more established specifications whose testing or examination verifies that the production and process control system produces a supplement meeting all applicable product specifications. Under 111.75(c)(2), appropriate tests or examinations must determine compliance with those selected specifications.

111.75(c)(3) requires an adequately documented basis explaining why compliance with the selected specifications ensures the finished batch meets all product specifications. Quality control personnel must review and approve that documentation under 111.75(c)(4); assign this review explicitly within the quality control unit’s responsibilities.

Connect upstream evidence to that rationale. 21 CFR 111.70(c)(2) requires documenting why meeting in-process specifications, together with component specifications, helps ensure finished-product attributes and contamination limits are met. Component and process results should therefore support an explained assurance chain, not sit as unrelated attachments.

Separate batch selection from sample collection

Batch selection identifies which batches enter verification. Within-batch sampling addresses where and how material is collected. 21 CFR 111.80(c) calls for representative samples of the selected finished batches before release for distribution.

As practical plan fields—not a prescribed sampling formula—document:

• Population: products, batch types and period covered.
• Selection rule: how batches enter the subset and the statistical justification.
• Variability assumptions: supporting process history and limits of that evidence.
• Collection design: sample locations, timing, quantities and handling.
• Review triggers: failures, trends or changes routed through change control.

FAQ: Can an attribute be exempted when no finished-batch method exists?

Only conditionally. Under 21 CFR 111.75(d)(1), you must determine and document both that the selected specifications cannot verify the exempted product specification and that no scientifically valid method exists for testing or examining it at the finished-batch stage. You must also document why alternative evidence—such as component and in-process testing, examination or monitoring—will ensure that specification is met without periodic finished-batch testing. Under 111.75(d)(2), quality control personnel must review and approve that documentation.

Use one traceable lifecycle for each dietary supplement batch. As an operating practice, identify the applicable approved specification revision, connect component and in-process control evidence, and identify batches selected for finished-product verification under the approach covered above. Keep those decisions attached to the batch rather than reconstructing them at release.

21 CFR 111.80(c) requires representative samples from finished batches selected through a sound statistical sampling plan, or otherwise every finished batch, before release for distribution to verify conformity with product specifications. Operationally, connect each sample to its batch, applicable specification revision and resulting laboratory report.

After results arrive, reconcile them with upstream controls and unresolved exceptions. Recommended result states are entered, reviewed and accepted, with named reviewers: entered means recorded; reviewed means checked; accepted means judged suitable for the conformity assessment. None alone means the batch has been dispositioned.

21 CFR 111.123(a)(7) places determination of each finished batch’s conformity with quality control. Under 111.123(a)(8), QC approves and releases, or rejects, each finished batch for distribution, including reprocessed batches. Record that human batch release decision separately from result acceptance; a populated results screen is not a release decision.

Who decides whether reprocessing is acceptable?

QC approves or rejects reprocessing under 21 CFR 111.123(a)(5). Apply the batch-conformity prohibition discussed in the failure-controls section; reprocessing approval is not distribution approval.

Under 21 CFR 111.260(h), the batch record must include testing or examination results from batch production, or cross-references to them. 21 CFR 111.260(i) also requires documentation that the finished dietary supplement meets specifications established under § 111.70(e) and (g). A report reference and a documented conformity assessment serve different purposes.

Link evidence to the intended criterion

As vendor-neutral practice, join the applicable specification revision, criterion, verification plan, sample, laboratory result and batch review. These are proposed logical records, not a named application's internal schema. Each result should resolve to the criterion used for its assessment.

For laboratory, manufacturing and document-system interfaces, use MES-to-LIMS integration checks to match product, batch and sample identifiers. Retain method and specification versions, preserve original units and reports, and identify corrected or superseded results. Reconcile identifiers at sample login and handoff, not only during final review.

An illustrative external-laboratory cross-reference is: report EXT-042; laboratory Example Analytical; sample SMP-017; revision 2; durable retrieval location: controlled document repository, record LAB/EXT-042/rev2. Test that reviewers can retrieve that exact revision. An unexplained attachment filename such as “final.pdf” does not provide the same traceable access.

Recover without erasing history

Route a wrong sample identifier, missing reporting basis or conflicting report revision to a clarification queue. Retain the original import, obtain an attributable correction, re-evaluate against the intended criterion and document reviewer resolution. Do not silently overwrite the result or treat an unresolved import as accepted evidence.

Proposed evidence model: Specification revision → criterion → verification plan → sample → laboratory result → batch review.

Use these practical failure-and-control pairs to challenge a finished-product specification before relying on it for batch review. They are preventive checks, not a ranking of inspection findings.

The binding release boundaries are separate from those preventive controls. Under 21 CFR 111.123(b)(1), quality control personnel must not approve and release for distribution any dietary supplement batch containing a component that fails its identity specification. Under 21 CFR 111.123(b)(2), they must not approve and release any finished batch, including a reprocessed batch, that does not meet all product specifications established under 111.70(e). Apply these boundaries during batch release.

As an evidence-integrity control in out-of-specification handling, preserve failed and corrected results with their investigation context. A later passing result should not erase earlier evidence. Keep the reason for correction and the reviewer’s disposition visible; this recommendation does not prescribe a retesting rule.

Use evidence retrieval as a procurement test, not a feature checklist. Ask a reviewer to identify the applicable specification revision, trace a sample to its laboratory result, explain the verification rationale and locate the disposition record.

Ask vendors to demonstrate four evaluation scenarios: a corrected laboratory result, an incompatible reporting unit, a superseded specification and an unresolved batch exception. Assess what remains visible and how discrepancies reach reviewers.

Distinguish laboratory results, manufacturing execution records and controlled specifications. Establish an authoritative source for each record and a reconciliation process across the laboratory system, manufacturing record system and controlled-document repository. Make these boundaries explicit in MES–LIMS integration evaluations.

What should contract-manufacturing agreements address?

For contract manufacturing, document who proposes specifications, maintains methods, supplies results, communicates changes and performs assigned reviews. Do not infer legal responsibility from brand-owner or contractor labels alone.

How should supplier certificates be evaluated?

Identify the characteristic, material and lot each certificate addresses, then map its contribution to the finished-product evidence strategy. These are procurement recommendations, not certificate-reliance permissions.

Use this recommended implementation exercise for one dietary supplement. Assign named owners to the roles below and record document IDs, revisions and locations—not verbal assurances.

Using the distinctions covered above, walk one completed batch record from product specification through sample collection, test results, upstream component and process evidence, and final review. Ask an unfamiliar reviewer to follow every reference. Log each point needing oral explanation as a traceability gap, with an owner and corrective action.

Repeat with a corrected laboratory report or changed ingredient source. Test whether superseded records remain distinguishable, affected decisions are reassessed, and recovery preserves the evidence chain—not merely whether the clean path works.

Use change control to trigger review after formulation, ingredient-source or analytical-method changes, changed intake instructions, recurring process variation, or new contamination information. Set review timing by impact; this checklist adds no mandatory calendar interval.