Tell us what products you need, your target market, estimated quantities, required specifications, packaging requirements, and any documentation your company requires.
Still Looking for a Reliable Disposable PPE Supplier?
Tell us what disposable PPE you need, including specifications, quantity, target market, and OEM needs. We’ll review it and recommend suitable options.
Contact Our Team
Tell us what products you need, your target market, estimated quantities, required specifications, packaging requirements, and any documentation your company requires.
Have a PPE Product Requirement ? Talk to Our Specialists
Get practical guidance on disposable PPE specifications, materials, sizing, packaging, private labeling, documentation, and bulk production requirements.
A test report only protects you if the sample behind it was chosen properly. Buyers, auditors, and regulators often accept a passing result without asking how many masks were pulled, from where, and by whom. When a shipment is disputed, those are the first questions anyone asks.
A defensible sampling plan is one you can explain, repeat, and justify with a statistical basis. This guide covers how to build one for medical masks, from defining the lot to documenting the results.
1. Start With the Standard Your Mask Claims
Sampling must serve a specific performance claim. Define the claim first, because it determines which tests you run and how many specimens each one needs.
EN 14683 (Types I, II, IIR): covers BFE, differential pressure (breathability), microbial cleanliness (bioburden), and splash resistance for Type IIR.
YY 0469 / YY/T 0969: the Chinese standards for surgical and disposable medical masks.
Respirators (N95, FFP2, KN95): these fall under different standards and regulators, such as NIOSH 42 CFR 84, EN 149, and GB 2626. They have their own conformity and batch-testing requirements and should not be treated as surgical masks.
Standards usually specify how many specimens to test for each method. Some, such as ASTM F2100, also reference an acceptance quality limit (AQL) for certain performance tests. Always check the current edition. Your plan should extend those minimums, not replace them.
2. Define the Lot Before You Sample It
Most sampling failures start with a badly defined lot. A lot should be a group of masks that are as uniform as possible. In practice that means:
The same product model and specification
The same production line and period
The same raw material lots, especially the meltblown filter layer
The same nose wire, ear loop, and sterilization batch (if applicable)
If you mix lots in one sample, a pass tells you little about any of them. Cartons from different meltblown rolls or different shifts should be treated as separate lots, or stratified inside one lot (see section 5).
Record the lot size, production dates, and material lot numbers on the sampling record. Without that traceability, a passing result is hard to defend.
3. Separate Defect Classes: Critical, Major, Minor
Not every defect carries the same risk, so one blanket AQL isn’t defensible. Split inspection into classes:
Class
Examples
Typical approach
Critical
Missing filter layer, holes, foreign matter, wrong marking of mask type or level
AQL not applicable, or zero tolerance
Major
Broken ear loops, weak nose wire, bad welds, wrong dimensions, missing labels
Tight AQL (e.g., 1.5 or 2.5)
Minor
Cosmetic marks, slight misprints, uneven pleats
Looser AQL (e.g., 4.0)
These AQL values are common industry practice, not regulatory requirements. Set them in your quality agreement with the supplier or in your internal specification, and apply them consistently.
4. Choose a Sampling Scheme and Calculate Sample Size
Attribute (visual and dimensional) inspection
ISO 2859-1, which corresponds to ANSI/ASQ Z1.4, is the usual framework. You pick an inspection level and an AQL, and the tables give a sample-size code letter and an accept/reject number.
Some commonly used sample sizes:
Code letter
Sample size
D
8
E
13
F
20
G
32
H
50
J
80
K
125
L
200
M
315
General Level II is the usual default for visual inspection.
Special levels (S-1 to S-4) suit destructive or expensive tests such as BFE or fluid resistance, where large samples aren’t practical.
Tightened or reduced inspection (the switching rules) lets you respond to supplier history. Tighten after repeated rejections and consider reducing after a sustained clean record.
Zero-acceptance plans for critical issues
When any failure is unacceptable, use a c = 0 plan. The sample size depends on the defect rate you want to detect and your confidence level. At 95% confidence, the approximate size is 3 divided by the defect rate:
Defect rate you want to detect
Sample size (c = 0, 95% confidence)
10%
29
5%
59
2.5%
119
1%
299
This table shows the trade-off clearly. A small sample can only catch large problems. If your buyer expects assurance at a 1% defect rate, 20 masks won’t provide it.
Performance testing
Lab tests like BFE, PFE, differential pressure, and fluid resistance are destructive and costly. Follow the specimen counts in the standard. Then decide how many masks per lot go to the lab, and how many per layer or location. Many buyers test one set per lot for routine release. They add extra testing for a new supplier, a new material lot, or any process change.
Note the acceptance criteria carefully. For example, ASTM F2100 applies an AQL-based pass criterion to some tests, so a small number of failing specimens can still pass the lot. EN 14683 sets its own limits and conformity rules. Don’t assume the two work the same way.
5. Make the Sample Representative
A statistically correct sample size is worthless if the masks aren’t drawn randomly.
Draw from across the whole lot. Sample from different pallets, cartons, and positions (top, middle, bottom), not the most accessible cartons.
Use stratified sampling when a lot spans multiple shifts, lines, or material rolls. Draw proportionally from each stratum.
Use random selection. Random number tables or a simple generator work. Don’t let the supplier choose the cartons.
Sample from the sealed shipping unit. Open the cartons yourself or have a third party do it, and take masks from the inner packs.
Keep the samples intact. Seal, label, and retain them with a chain-of-custody record so nobody can question what the lab tested.
Also keep retention samples from each lot. If a complaint appears months later, you can retest the same masks.
6. Decide Who Samples and Who Tests
Independence strengthens your position. The strongest setup is a third-party inspector who samples on-site and sends the masks to an accredited lab (ISO/IEC 17025). If you can’t use a third party, document your internal procedure carefully and make sure the lab’s scope covers each method you use.
Check that the lab:
Is accredited for the specific test methods (BFE, PFE, delta P, fluid resistance)
Reports the mask conditioning, test flow rates, and specimen counts
Issues reports that identify the lot, sample size, and sampling date
7. Define Decision Rules Before You Test
Write the following down before the first mask is opened:
Accept/reject numbers for each defect class
What happens on rejection: 100% sorting, resampling, or a full lot return
Whether a retest is allowed, and under what conditions. Retesting until a lot passes is a classic credibility killer.
Who decides on borderline results
How a failed lot affects the switching rules for future shipments
8. Document Everything
Your sampling file for each lot should show:
Lot number, size, and production dates
The standard and acceptance criteria used
Inspection level, AQL, sample size, and accept/reject numbers
The sampling method (random, stratified) and who performed it
Carton and pallet locations of the sampled units
Inspection and lab results, with the raw data
The disposition (accepted, rejected, or reworked) and who signed off
If the paperwork is complete, you can reconstruct and defend every decision.
9. Apply the Same Logic Across Your Wider PPE Program
Many procurement teams buy masks alongside other disposables, and the same sampling framework carries across product lines. The attributes you inspect will differ (seam strength, fluid barrier, fit, sole adhesion), but the lot definition, AQL structure, random draw, and documentation rules stay the same.