Sampling Plan Compliance Requirements

Sampling plan compliance defines the structured statistical methods organizations must follow when inspecting lots, batches, or production runs to make accept/reject decisions without testing every unit. Federal agencies including the FDA and the Department of Defense, as well as voluntary standards bodies such as ANSI and ASQ, have codified specific sampling requirements that carry regulatory force across industries from medical devices to food processing to defense procurement. Understanding which plan type applies, how to construct it correctly, and where acceptance boundaries sit determines whether a product release decision will withstand an audit or enforcement review. This page covers the regulatory framing, operational mechanics, scenario-based applications, and the decision logic that separates compliant sampling from statistically indefensible inspection.

Definition and scope

A sampling plan is a documented procedure specifying the sample size drawn from a lot, the criteria for accepting or rejecting that lot, and the statistical risk parameters—typically expressed as the Acceptable Quality Level (AQL) and the Lot Tolerance Percent Defective (LTPD)—that govern the decision. The scope of sampling plan compliance extends to any regulated product or process where 100-percent inspection is impractical, destructive, or cost-prohibitive.

The primary codified framework in the United States is ANSI/ASQ Z1.4 for attribute sampling (pass/fail characteristics) and ANSI/ASQ Z1.9 for variables sampling (measurable continuous characteristics). These standards were derived from the military standard MIL-STD-1916 and the earlier MIL-STD-105E, and they remain the accepted reference documents cited by FDA guidance, USDA program requirements, and third-party certification schemes. FDA's Quality System Regulation at 21 CFR Part 820 requires that sampling procedures be based on a valid statistical rationale—a requirement reinforced in the updated Quality Management System Regulation finalized in 2024 under 21 CFR Part 820.250.

Compliance scope also depends on industry sector. Food manufacturers operating under 21 CFR Part 110 (Good Manufacturing Practice) and defense contractors bound by MIL-STD-1916 face different but overlapping documentation and traceability obligations when sampling plans are challenged. For a broader view of how sampling sits within a larger inspection and testing framework, see Inspection and Testing Compliance.

How it works

A compliant sampling plan is constructed and executed in four discrete phases:

  1. Lot definition and stratification. The lot must be defined before sampling begins—by production date, batch number, supplier shipment, or process run. Undefined lot boundaries are a primary audit finding because they prevent traceability back to a root cause if a lot is later rejected. See Traceability Requirements Quality Control for the documentation obligations that attach to lot definition.

  2. AQL and inspection level selection. The AQL is chosen based on the severity of the characteristic being inspected—typically 0.065 to 1.0 for critical defects, 1.0 to 2.5 for major defects, and 4.0 for minor defects under ANSI/ASQ Z1.4 conventions. The inspection level (I, II, or III) determines the sensitivity of the plan; General Inspection Level II is the default in most regulatory contexts unless a specific risk assessment justifies a different level.

  3. Sample size and acceptance number determination. Using the AQL and lot size, the practitioner references the appropriate table in Z1.4 or Z1.9 to identify the sample size code letter and corresponding acceptance number (Ac) and rejection number (Re). For example, a lot of 1,200 units inspected at General Level II with an AQL of 1.0 yields a sample size of 80 units, with Ac = 2 and Re = 3.

  4. Switching rules and plan type. ANSI/ASQ Z1.4 specifies three plan types—normal, tightened, and reduced—with mandatory switching rules. A producer shifts to tightened inspection after 2 consecutive lots are rejected; 5 consecutive accepted lots on tightened inspection allows a return to normal. Ignoring these switching rules constitutes a compliance gap independent of the accept/reject outcome for any single lot.

Common scenarios

Incoming supplier material. Manufacturers receiving raw materials or components apply attribute sampling plans at receiving inspection. FDA-regulated device manufacturers must document the AQL rationale as part of supplier controls under 21 CFR Part 820.50. The Supplier Quality Compliance framework details how incoming sampling integrates with approved supplier programs.

In-process inspection. Statistical process control checkpoints during production may use reduced sampling plans only when process stability has been demonstrated over a defined prior history. ANSI/ASQ Z1.9 is particularly applicable here because many in-process characteristics (dimensions, weights, fill volumes) are continuous variables for which variables plans are statistically more efficient—requiring roughly 40 percent fewer samples than attribute plans for the same risk protection, as noted in ASQ's published comparison tables.

Final product release. Pre-shipment sampling for final release is the most scrutinized application. In pharmaceutical and medical device contexts, 21 CFR Part 211.165 requires laboratory testing on a representative sample from each batch, with the sampling procedure specifying the sample size and the basis for the statistical confidence it provides.

Destructive or costly testing. When testing destroys the unit (tensile strength, burst pressure, sterility), reduced or skip-lot plans governed by ANSI/ASQ Z1.4 Appendix B may apply, provided the process meets documented stability prerequisites.

Decision boundaries

The central compliance decision boundary is the accept/reject threshold, but two additional boundaries carry equal regulatory weight:

Attribute vs. variables plan selection. Attribute plans (Z1.4) classify each unit as conforming or nonconforming. Variables plans (Z1.9) use the mean and standard deviation of the sample to estimate the fraction nonconforming. Variables plans produce tighter confidence intervals for the same sample size but require that the characteristic follow a normal distribution—an assumption that must be validated and documented. Selecting a variables plan for a non-normal distribution without transformation or justification is an audit failure point.

Consumer risk (β) vs. producer risk (α). The AQL governs producer risk—the probability that a good lot will be rejected. The LTPD or Rejectable Quality Level (RQL) governs consumer risk—the probability that a bad lot will be accepted. At the standard 10 percent consumer risk point, lot tolerance percent defective values range from approximately 3 to 10 times the AQL depending on sample size. Regulatory reviewers and auditors expect practitioners to state both risk parameters explicitly, not just the AQL. Documenting only the AQL without the associated consumer risk is a common nonconformance finding under quality-control-compliance-requirements.

When sampling results yield a borderline condition—defects equal to the acceptance number—the lot is accepted, not held for additional review, unless an independent risk-based review process has been pre-documented. Improvised hold decisions after a borderline acceptance result, without a pre-established procedure, expose organizations to findings under both FDA quality system regulations and ISO 9001:2015 Clause 8.6 (release of products and services). Connecting sampling outcomes to corrective action triggers is addressed in Corrective and Preventive Action Compliance.

References

📜 1 regulatory citation referenced  ·  🔍 Monitored by ANA Regulatory Watch  ·  View update log

Explore This Site

Services & Options Compliance: Standards Overview Regulations & Safety Regulatory References Tools & Calculators Contractor License Fee Calculator