How Many Units to Inspect on Receipt: AQL Sampling for Sterile Consumables

What changed in ISO 2859-1 in January 2026 — and does your quality agreement cite the right edition?

Every week, international medical device distributors, contract manufacturing buyers, and hospital central-supply quality managers receive containerized shipments of terminally sterilized consumables from Chinese manufacturing facilities. These consignments routinely combine diverse product families: disposable circumcision devices, wound-care negative pressure dressing kits, sterile injection needles, and puncture assemblies. For the receiving quality engineer, the immediate dilemma is operational: exactly how many sealed cartons and sterile pouches must be pulled, opened, and evaluated before signing the incoming acceptance log?

In consumer goods and apparel, incoming inspection relies on online Acceptable Quality Limit (AQL) calculators that assume unlimited destructive or non-destructive handling of cheap, non-sterile articles. In sterile medical devices, that assumption breaks down instantly. Ripping open a sterile Tyvek-poly pouch to inspect an internal component immediately destroys the sterile barrier, voids the remaining shelf life, and renders the unit unsaleable clinical waste. Furthermore, if a distributor QA team opens twenty pouches to perform dye-leak testing, who pays for the destroyed units?

The international benchmark for attribute acceptance sampling is ISO 2859-1. ISO's catalog dates the third edition, ISO 2859-1:2026 (Sampling procedures for inspection by attributes — Part 1: Sampling schemes indexed by acceptance quality limit (AQL) for lot-by-lot inspection), to January 2026.¹ The second edition, ISO 2859-1:1999, was withdrawn on 22 January 2026.³ The catalog field itself is month-level. The day used here is the recorded withdrawal date of the edition it replaced.

Quality agreements should pin the edition they mean. The fear that every sample size and accept/reject number moved in January 2026 is not what the published change list says. The ISO Online Browsing Platform foreword lists four changes: a procedure for switching from normal or reduced inspection to skip-lot sampling; guidance on producer and product qualification for that switch; guidance on skip-lot efficacy and on choosing lots at random; and replacement of the printed per-plan operating-characteristic curves with methods, now in Annex E, for constructing an individual plan's OC and average-sample-number curves.² The third edition also folds in Amendment 1:2011 and Technical Corrigendum 1:2001. That list does not announce a new AQL series, new code letters, or new accept/reject numbers. The full 2026 tables are not on the public preview, so the worked numbers in this guide are read from the public 1999 text (IS 2500-1, the Indian adoption of ISO 2859-1:1999) and labelled as such.⁴ Check the licensed 2026 tables before you treat those numbers as a cell-by-cell proof of the new edition.

What the 2026 foreword actually adds:

• Skip-lot switching: the foreword adds a procedure for moving from normal or reduced inspection to skip-lot sampling. In the 1999 text, clause 9.5 only pointed to ISO 2859-3. The 2026 scope preview places optional skip-lot procedures in clause 9.5 of this part.
• Qualification guidance: the foreword says guidance on producer and product qualification has been added. The public preview does not print a capability index, a production-month minimum, an ISO 13485 certificate test, or a ten-lot counter.
• Random selection and efficacy: the foreword adds guidance on how well a skip-lot program is working and on methods for choosing, at random, which lots to inspect and which to skip.
• Annex E: the per-plan operating-characteristic curves are removed. Methods to construct an individual plan's OC curve and its average sample number curve are placed in Annex E.
• Prior corrections: the third edition incorporates the 2011 amendment and the 2001 corrigendum, so a contract that cited only the unamended 1999 text was already behind those corrections.

Current-Law Anchor: FDA QMSR and Reserved 21 CFR 820.250

A frequent misconception encountered across vendor blogs and older quality manuals is that the US FDA explicitly mandates a numerical sampling plan or an AQL threshold (such as 'FDA requires AQL 1.0 for medical packaging'). No regulator mandates a specific numerical AQL for incoming device inspection.

Furthermore, medical device quality managers must be vigilant regarding stale regulatory citations. On 2 February 2026, the FDA's Quality Management System Regulation (QMSR) final rule took full effect, aligning Title 21 of the Code of Federal Regulations with ISO 13485:2016⁶. Under the revised 21 CFR Part 820, Subparts C through O are officially reserved⁷. This means that 21 CFR 820.80 ('Acceptance activities') and 21 CFR 820.250 ('Statistical techniques') no longer exist as enforceable regulatory text. Articles or vendor questionnaires quoting 21 CFR 820.250(b)—'Sampling plans, when used, shall be written and based on a valid statistical rationale'—are citing a retired regulation⁹¹⁰.

Under current US device CGMP, acceptance obligations are the ISO 13485:2016 clauses that QMSR incorporates by reference. FDA inspects manufacturers under Compliance Program 7382.850.⁵⁶ The clauses that carry the duty are:

• Clause 7.4.3 (Verification of purchased product): 'The organization shall establish and implement the inspection or other activities necessary for ensuring that purchased product meets specified purchase requirements.'
• Clause 8.1 (General): determine suitable methods, including statistical techniques, and the extent of their use. That is the statistical-rationale duty older pages still pin on the reserved text of 21 CFR 820.250(b).
• Clause 8.2.6 (Monitoring and measurement of product): monitor and measure product characteristics against the requirements, keep evidence of conformity with the acceptance criteria, record who authorized release, and do not release product until the planned and documented arrangements are complete. ISO 13485:2016 has no clause 8.6. Release is inside 8.2.6. Clause 8.6 is the ISO 9001:2015 title for the counterpart requirement.

Regulators do not mandate what your AQL numbers must be; they mandate that you establish an objective, statistically defensible procedure, document your rationale, verify conformity against defined risk files, and record lot disposition before releasing product into commercial distribution. ISO 2859-1:2026 provides the recognized statistical methodology to satisfy ISO 13485:2016 Clause 7.4.3 and 8.2.6.

How do you define the inspection lot when one container carries three SKUs?

The receiving error that breaks the plan happens before anyone opens a table. A supervisor treats every unit in the container as one inspection lot. For mixed sterile consumables, that lot is not what ISO 2859-1 asks you to form.

Two different sentences in the public 1999 text are easy to collapse.⁴ Clause 3.1.13 defines a lot as a definite amount of product, material or service collected together. Its note says an inspection lot may consist of several batches or parts of batches. Clause 6.1 is the formation rule: as far as is practicable, each lot shall consist of items of a single type, grade, class, size and composition, manufactured under uniform conditions at essentially the same time. The 2026 change list does not announce a revision of those clauses. The long sentence is therefore not 'clause 3.1,' and it is not a rule that one batch number is always one lot. For a terminally sterilized device, batches that do not share the sterilization cycle and the same process conditions fail the 'uniform conditions at essentially the same time' test, so split them before you read a code letter. Batches of the same SKU that really were made together can sit in one inspection lot. The standard says 'as far as is practicable,' not 'one carton tally, one sample.'

Worked Container Scenario: 20,000 Units across Three Product Families

Consider an international distributor receiving a single consolidated shipping container containing 20,000 total devices from Weihai Medison Medical Equipment Co., Ltd. (VEMERIX). The commercial packing list reveals the following distribution:

• Disposable Circumcision Rings (adult surgical platform): 4,000 units split across two manufacturing batches: Lot CR-260801 (2,500 units) and Lot CR-260802 (1,500 units).
• NPWT Dressing Kits (sterile polyurethane foam with drape): 6,000 units from a single manufacturing batch: Lot NW-260810 (6,000 units).
• Sterile injection needles: 10,000 units split across two manufacturing batches: Lot IN-260815 (5,000 units) and Lot IN-260816 (5,000 units).

If the receiving team mistakenly aggregated the container as a 20,000-unit lot under General Inspection Level II, Table 1 would yield Code Letter M, requiring 315 units sampled. But how would those 315 units be distributed? If all 315 were drawn from the easiest-to-reach pallets of injection needles, the circumcision rings and NPWT dressings would have received exactly zero physical verification.

On this packing list the five batch numbers are five inspection lots. They are not one type manufactured under one set of conditions, so clause 6.1 does not let you draw a single code letter for the container:

Before drawing a single physical unit from any pallet, the receiving quality lead executes Document & Lot Reconciliation. The shipping invoice, packing list, bill of lading, and case barcode labels must be matched against the supplier's Certificate of Analysis (CoA) pack. If Lot CR-260801 is missing its batch sterilization record or shows a quantity mismatch against the carton tally, the lot is placed on immediate documentary hold without wasting technician labor on physical sampling.

Which inspection level applies — General II, or special levels for destructive checks?

ISO 2859-1 establishes two distinct tiers of inspection levels: General Inspection Levels (I, II, III) and Special Inspection Levels (S-1, S-2, S-3, S-4). Selecting between them is not an arbitrary cost-saving preference; it reflects the physical nature of the test and the statistical discrimination required.

General Inspection Level II is the default unless the responsible authority specifies otherwise. That is clause 10.1 of the public 1999 text, and the 2026 change list does not announce a different default.⁴ Level I and Level III are not a fixed percentage of Level II. For a lot of 3,201 to 10,000 units, Table 1 gives General I as code J (sample size 80), General II as code L (200) and General III as code M (315). Level I is two code letters below Level II in that bracket, not one, and 80 is 60 percent smaller than 200. Level III is the next code letter up, 315 rather than 200. Read the row for the lot in front of you. Do not apply a '40 percent' rule from a different bracket.

Special Levels S-1 through S-4 exist explicitly for tests that involve high unit costs, destructive sample consumption, or lengthy testing durations. For a 5,000-unit consumable batch, General Level II demands 200 units. For a non-destructive barcode scan or outer pouch visual check, 200 units can be surveyed in 45 minutes without compromising the devices. However, if an engineer subjected 200 sterile circumcision staplers or 200 complex NPWT suction canisters to destructive burst or tensile pull testing, the testing process itself would destroy thousands of dollars in commercial stock.

Operating Characteristics and the High-Cost Inspection Tradeoff

Moving from General Level II to Special Level S-3 or S-2 costs discrimination. For a lot of 3,201 to 10,000, S-3 is code F (n = 20) and S-2 is code D (n = 8).⁴ At General II the normal single-sampling plan for AQL 1.0 is n = 200, Ac 5, Re 6. At a 3 percent nonconforming rate, the chance of rejecting that lot is about 56 percent (binomial: n = 200, accept on 5 or fewer, p = 0.03). The curve is steeper than a 20-unit plan. It is not a near-certain screen.

Under S-3 (n = 20), a lot with 2 percent defective seals has about a 67 percent chance of showing zero defects in the sample (0.98²⁰). The special-level sample is a spot check of a validated process. It is not the control that keeps a low seal-defect rate out of the warehouse.

How do you set AQLs by defect class for sterile single-use devices?

A pervasive error in procurement negotiations is interpreting the AQL as an acceptable defect allowance. Quality leads frequently hear commercial staff say, 'Our quality agreement specifies AQL 2.5, which means the factory is allowed to send us 2.5% bad product.' That is not what clause 3.1.26 says. A June 2022 manufacturing-software article still defines AQL as the maximum number of defects allowed in a lot before the lot is rejected.⁹

In clause 3.1.26 of the public 1999 text, the acceptance quality limit (AQL) is the quality level that is the worst tolerable process average when a continuing series of lots is submitted for acceptance sampling.⁴ It indexes a sampling plan. It is not a permit to ship that percent defective. The same clause notes that a lot as bad as the AQL may still be accepted with fairly high probability, and that naming an AQL does not make that level desirable. If the process average is not better than the AQL, the chance of a switch to tightened inspection is high. Once there, discontinuation of sampling becomes likely unless the process improves. That discontinuation pauses acceptance under the standard. It is not an automatic ban on the supplier's shipments.

In medical devices, attributes are segregated into three standardized defect classes based on clinical and regulatory severity:

• Critical Defects: Nonconformities that entirely compromise the sterile barrier, create direct patient safety hazards, or violate statutory labeling mandates. Examples: open pouch seal channels, punctures or tears in Tyvek, missing ethylene oxide sterilization indicator transition, unreadable lot/expiry dates, or foreign particulate inside the sterile fluid path.
• Major Defects: Nonconformities that impair device functionality, increase procedure time, or create packaging handling failures without breaching the sterile barrier. Examples: seal width narrowing below specification (without a channel), deformed blister tray flange, bent needle hub preventing luer connection, or illegible secondary carton typography.
• Minor Defects: Cosmetic or packaging anomalies that do not affect device performance, safety, or sterility. Examples: scuffed outer shipper cartons, slightly misaligned tape, or non-critical labeling typography blemishes.

The Zero-Acceptance Convention (Ac 0 / Re 1) for Critical Barrier Breaches

For critical defects, particularly sterile barrier breaches, modern medical device quality systems enforce a strict Zero-Acceptance rule. In procurement contracts and published site guides—such as the mesotherapy injector procurement acceptance guide—this is frequently expressed colloquially as 'AQL 0.0 for Critical Defects'.

The ISO 2859-1 AQL series starts at 0.010. A literal AQL of 0 is not a column in the tables.⁴ Contracts that say 'AQL 0 for critical defects' usually mean a separate Ac = 0, Re = 1 plan: one critical defect in the sample rejects the lot. That is a buyer convention. It is not how the standard defines an AQL, and it is not the same plan as AQL 0.65. Quarantine follows if the quality agreement says the lot is rejected. The standard does not itself order a warehouse hold.

These plans do not all say the same thing, and they should not be blended. The disposable circumcision device bulk-price guide uses a contractual AQL of 0.65 for sterile-barrier integrity and AQL 1.0 for mechanical function. At General II, AQL 0.65 is not Ac 0. On the public 1999 normal single-sampling table it is Ac 2 / Re 3 at code K (n = 125), Ac 3 / Re 4 at code L (n = 200), and Ac 5 / Re 6 at code M (n = 315).⁴ AQL 1.0 at those same letters is the major-defect column in the table above. The sterile hypodermic needle acceptance guide leaves the sample size to the buyer's quality agreement. The mesotherapy injector guide's phrase 'AQL 0.0 for critical defects' is the colloquial Ac 0 / Re 1 convention, not an indexed AQL. Pick one plan per defect class and write the code letter, n, Ac and Re into the quality agreement.

What does opening sample units cost you — and what does the CoA replace?

The fundamental tension in receiving sterile single-use devices is that attribute sampling consumes the very properties being verified. This physical reality forces a rigorous division between physical dock inspection and documentary lot reliance. Quality engineers call this the 'Sterile Split'.

Why Receiving-Dock Sterility Testing Is a Dangerous Fallacy

Occasionally, an uninformed procurement committee asks: 'Can our hospital laboratory perform sterility testing on 20 incoming pouches to verify the lot is sterile?' In medical device quality assurance, sterility testing on receiving inspection is a recognized fallacy.

Under pharmacopeial standards (USP <71> / EP 2.6.1) and ISO 11737-2, sterility testing involves transferring devices into fluid thioglycollate medium and soybean-casein digest medium, followed by 14 days of incubation. Beyond the 14-day delay and testing cost, the statistical reality is devastating: to achieve an accepted Sterility Assurance Level (SAL) of 10⁻⁶ (a one-in-a-million probability of a non-sterile unit), testing a sample of 20 or even 200 units has virtually zero statistical power to detect a low defect rate. An engineer could test 200 units from a batch with a 1-in-1,000 non-sterile rate, and there is about an 82 percent probability that all 200 units would test sterile (0.999²⁰⁰).

Sterility is not inspected into medical devices; it is manufactured in through validated terminal sterilization processes. Where the product page states ethylene oxide sterilization, as it does for the circumcision device, acceptance relies on documentary verification of the batch sterilization report against ISO 11135 and GB 18279-2023. Do not copy that EO line onto the NPWT dressing or the injection needle; those pages do not state an ethylene oxide cycle. As established in the EO sterilization CoA incoming guide, the receiving engineer verifies four critical parameters on the supplier's certificate:

• Chamber load: the incoming batch number appears on the loading record for the cycle being cited.
• Cycle parameters: gas concentration, temperature, relative humidity, dwell and pressure match the validated cycle. Humidity is the value that validation specified. ISO 11135 does not impose a universal floor of 50 percent relative humidity.
• Biological indicator: the certificate states the indicator result for the cycle, commonly no growth from Bacillus atrophaeus challenges placed in the load. Read the stated population and the incubation result. A negative indicator is lot evidence, not a count of every organism in the chamber.
• Residuals: read the contact category before you compare milligrams. Under the FDA-recognized ISO 10993-7:2008 text, 4 mg of ethylene oxide and 9 mg of ethylene chlorohydrin per device are the adult limited-contact ceilings (24 hours or less) under FDA Recognition Number 2-275. They are not a prolonged-contact rule and not an average daily dose. The ISO 10993-7 recognition guide explains why a 4 mg / 9 mg line on a 2026 header is not, by itself, the recognition position.

The Sterile Barrier Association's opening note is a handling rule, not a sampling plan: keep the device in its sterile barrier, store it clean and dry, avoid excess temperature, and inspect the barrier for damage before use. Opening the pack is a point-of-use step. It does not tell a receiving team how many units to pull.⁸

Disposition of a failed sterile lot is a quality-agreement question, not a second pass through the sterilizer by default. A buyer should not put a rejected single-use lot back on the shelf by sending the cartons through another ethylene oxide cycle. A repeat cycle can change residuals, materials and the aging evidence for the pack, and it needs its own validation before anyone treats the goods as released. The receiving action is quarantine and rejection. Replacement, credit or destruction belongs in the contract. The Sterile Barrier Association page cited here is about storage and opening, not about re-sterilization.⁸

What happens when a lot fails — switching rules and a rejected sterile lot

The true operational power of ISO 2859-1 lies not in isolated lot evaluation, but in its switching rules. ISO 2859-1 is designed as a continuous sampling system that actively penalizes erratic suppliers while rewarding mature, reliable manufacturing partners.

Incoming inspection operates under three primary inspection severities:

• Normal Inspection: The standard baseline applied when a procurement relationship begins or when quality history is unproven.
• Tightened Inspection: A severe sampling tier with stricter acceptance numbers (Ac) or smaller acceptance margins, designed to protect the buyer when a supplier's process average deteriorates.
• Reduced inspection: a smaller sample than the normal plan for the same code letter. Clause 9.3.3 allows it only when three things are true together: the switching score is at least 30, production is at a steady rate, and the responsible authority considers reduced inspection desirable. The score starts at zero on normal inspection. For single sampling, if the acceptance number is 2 or more, add 3 when the lot would still have been accepted at the next tighter AQL, otherwise reset the score to zero. If the acceptance number is 0 or 1, add 2 when the lot is accepted, otherwise reset to zero. It is not a count of ten accepted lots, and the reduced sample is not one-third of normal by rule.

Contractual Replacement, Credit, and Return Freight Safeguards

A rejected lot needs a contract remedy, or the quality finding stops at the quarantine cage. Nothing on the VEMERIX product pages sets a notice period, a freight term, or a destruction duty. Those are fields to negotiate. A workable quality agreement usually names three mechanics:

• Quarantine and notice: hold the lot and send the batch numbers, photographs and the sampling record within the period the quality agreement actually states.
• Supplier Investigation & Counter-Testing: The manufacturer reviews retained counter-samples from the same sterilization run at its cleanroom facility.
• Remedy: credit, replacement, return freight and disposal are negotiated terms. An example such as replacement within 14 days, or air freight at the supplier's cost, is a clause you may write. It is not a term published on the product pages.

What is skip-lot inspection in the 2026 edition, and when can a supplier earn it?

Skip-lot sampling is the part of the January 2026 edition that is actually new, and it is narrower than a vendor-tiering poster. In the public 1999 text, clause 9.5 says lot-by-lot inspection may be replaced by skip-lot sampling when ISO 2859-3 is met. The 2026 scope preview says this document itself specifies optional skip-lot procedures, and it points to clause 9.5.²⁴

The preview states the purpose and the mechanism. The purpose is to reduce inspection effort on product of high quality submitted by a producer who has a satisfactory quality-assurance system and effective quality controls. The mechanism is to decide at random, with a specified probability, whether a lot presented for inspection will be accepted without inspection. That extends the random selection of sample items to the random selection of lots. A fraction such as one lot in three is an illustration of a probability a buyer might write down. It is not a fraction printed in the public preview. Lots that are accepted without inspection are accepted without inspection. Keeping the sterilization certificate for every lot, including lots you do not open, is this guide's rule for sterile devices. It is not a sentence in the clause 9.5 preview.

For sterile single-use devices, keep the documentary gate even when physical sampling is skipped. Every manufacturing lot should still arrive with the certificate of analysis, the sterilization-cycle record and the certificate of conformance that the quality agreement requires. If a lot's file is missing, do not treat that lot as accepted without inspection, and stop using skip-lot until the file and the qualification are rebuilt. That boundary is a receiving rule for this class of product. The 2026 preview does not say that a missing certificate 'revokes' a clause.

The Receiving Inspection RFQ & Quality Agreement Clause Checklist

To ensure sampling protocols operate seamlessly across international supply chains, procurement leads must incorporate unambiguous acceptance clauses into their Request for Quotation (RFQ) documentation and bilateral Quality Agreements before placing purchase orders. Use the following structured checklist:

• Standard Edition Pinning: Explicitly cite 'ISO 2859-1:2026, Single Sampling Plans for Normal Inspection.' Specify that references to withdrawn editions (ISO 2859-1:1999) or retired regulations (21 CFR 820.250) are superseded.
• Inspection lot: state that the lot follows the formation rule — one type, grade, class, size and composition, made under uniform conditions at essentially the same time — and not the shipping container. Ask for batch identity on the pallet and on the packing list so sterilization loads that were not made together can be split.
• Defect classes: write the plan, not a slogan. For sterile-barrier breaches, say whether you mean Ac 0 / Re 1 or an indexed column such as AQL 0.65 (Ac 3 / Re 4 at n = 200 on the public 1999 normal table). For other majors, AQL 1.0, and for minors, AQL 2.5, copy the Ac and Re from the code letter rather than from memory.
• Destructive Sample Budget & Compensation: Establish that destructive testing (seal peel strength ASTM F88, dye penetration ASTM F1929, functional pull) operates under Special Inspection Level S-2 or S-3. Define whether the supplier provides an extra 0.5% overage units per shipment to cover destructive testing without eroding saleable stock.
• Mandatory Batch Documentation Pack: Require that every lot arriving at the dock be accompanied by: (1) Certificate of Analysis (CoA) with batch physical specs; (2) Batch Sterilization Cycle Report detailing EO parameters and biological indicator clearance; (3) Certificate of Conformance (CoC); and (4) Finished goods release checklist.
• Switching Rules & Penalty Escalation: Formally incorporate ISO 2859-1 switching rules. Stipulate that two lots not accepted out of five or fewer on original inspection switch the factory to tightened inspection under clause 9.3.1. Who pays for the extra sampling is a contract term, not a sentence in the standard.
• Rejection, Quarantine & Return Freight: Specify that rejected lots will be quarantined immediately. State the replacement window, who pays freight and duty, and who pays for destruction or return. Those are negotiated terms, not a published VEMERIX schedule.

Where VEMERIX fits — and where due diligence still begins

VEMERIX is the international brand of Weihai Medison Medical Equipment Co., Ltd., positioned as a Minimally Invasive Surgery Total Solution Platform serving urology, vascular surgery and perioperative care. This sampling plan is a buyer procedure. It is not evidence that a VEMERIX lot will pass it.

Product-page facts that matter when you write the lot definition, and no further:

• Circumcision device: the product page records an NMPA Class II registration (Lu Mech Reg. 20172020238), ethylene oxide sterilization, individual sterile blister packaging, and CE. See the circumcision device page. Do not copy that CE line onto other SKUs.
• Laser fiber: Medison's NMPA-registered single-use medical laser fiber (Lu Mech Reg. 20192010517) is used with a partner-supplied diode laser source. The fiber page does not make VEMERIX the manufacturer of that source. See the laser fiber page.
• NPWT dressing: the dressing kit page records a Class II kit. It does not state an EU MDR certificate or an ethylene oxide cycle. Ask for the lot sterilization record rather than assuming the circumcision device's EO line covers the dressing.
• Injection needle: the sterile single-use injection needle page records an NMPA Class III device in a sterile single-use format. It does not state a gauge, a coating, or a puncture-force value.

Ask for the lot file you will actually check: the sterilization record, the release record, and the sampling plan the factory used. A manufacturer's certificate does not replace the purchaser's own verification of purchased product under ISO 13485:2016 clause 7.4.3.⁵ The next step is a sample, a technical pack, or a distributor discussion through contact, not a claim that every consignment will pass.

Frequently Asked Questions

Is there a regulator-mandated AQL for incoming medical device inspection?

No. This guide did not find a US or EU rule that sets a numerical incoming AQL for medical devices. Under QMSR, effective 2 February 2026, 21 CFR 820.80 and 820.250 are reserved.⁷ The incorporated ISO 13485:2016 duties are clause 7.4.3 for purchased product and clause 8.2.6 for monitoring, measurement and release. Clause 8.1 expects suitable statistical techniques. The regulator does not pick your AQL column. The production and process-control file guide walks through why questionnaires that still cite 820.80 are citing a reserved section.

Does AQL 2.5 mean 2.5 percent defects are acceptable?

No. Clause 3.1.26 of the public 1999 text, which the 2026 change list does not announce as revised, defines the AQL as the worst tolerable process average when a continuing series of lots is submitted for acceptance sampling.⁴ A lot at that level may still be accepted with fairly high probability. The designation is not a permit to ship 2.5 percent defective, and it is not described as a desirable level. If the process average is not better than the AQL, tightened inspection becomes likely, and discontinuation of sampling becomes likely after that if the process is not improved.

What AQL should a sterile-barrier defect get?

Classify a sterile-barrier breach as critical in the defect list, then choose the plan on purpose. An Ac 0 / Re 1 plan rejects the lot on one such defect in the sample. That convention is common in contracts and it is not an AQL column. The indexed alternative used in the circumcision-device landed-cost guide, AQL 0.65, accepts some defects at these sample sizes: Ac 3 / Re 4 at n = 200, for example. Write which of the two plans you mean.

Do I have to open sterile packs to accept a lot?

No. Most incoming attribute checks should remain completely non-destructive. Outer carton integrity, shipper barcodes, secondary packaging, UDI labeling, and visual seal inspection (ASTM F1886) can be performed without breaching the sterile barrier. Destructive tests (ASTM F88 seal peel, ASTM F1929 dye penetration) should be budgeted on small samples using Special Inspection Levels (S-2 or S-3), while sterility assurance and EO residuals are verified via the supplier's certified batch sterilization records.

Should my quality agreement cite ISO 2859-1:2026 or the 1999 edition?

Cite ISO 2859-1:2026 if that is the edition you intend to follow. The catalog date is January 2026, and the 1999 edition was withdrawn on 22 January 2026.¹³ The published change list adds skip-lot procedures at clause 9.5 and moves curve construction to Annex E. It does not announce new sample sizes or accept/reject numbers. Keep the worked numbers in this guide traceable to the public 1999 tables until you have compared them with the licensed 2026 tables. Whether ANSI/ASQ Z1.4 has adopted the 2026 text was not established for this article.

How many units do I inspect for a 20,000-unit lot?

On the public 1999 General Inspection Level II single-sampling table, a 20,000-unit manufacturing lot falls in the 10,001 to 35,000 bracket, code letter M, sample size 315.⁴ The January 2026 change list does not announce a different code letter. If the quality agreement uses the plans in the worked table, critical defects on an Ac 0 / Re 1 convention reject on one defect; AQL 1.0 is Ac 7 / Re 8; AQL 2.5 is Ac 14 / Re 15. AQL 0.65 at this letter is Ac 5 / Re 6, not Ac 0.

Can I skip inspecting some lots from a long-standing supplier?

Only under the optional clause 9.5 procedures, and only after you have read the qualification guidance in the licensed text. The public preview requires a producer with a satisfactory quality-assurance system and effective quality controls, and it selects lots at random with a specified probability so that some lots are accepted without inspection.² It does not, in that preview, require an ISO 13485 certificate, a Cpk of 1.33, six months of production, or ten consecutive clean lots. For sterile devices, still file the lot sterilization record for every lot, including lots you do not open. That filing rule is a buyer control, not a line in the preview.

Does a sterilization CoA replace incoming sampling?

A sterilization CoA replaces biological and chemical testing (sterility incubation and EO residual gas chromatography), which cannot practically be performed at the receiving dock. It does not replace physical verification of shipping cartons, labeling, barcode scanning, or visual seal integrity. A valid receiving acceptance protocol combines documentary verification of the sterilization certificate with physical attribute sampling of packaging integrity.