Answers › c=0 sampling plans

What is a c=0 sampling plan, and how does it differ from an AQL plan?

Reviewed by the JCH tooling team · Updated

A c=0 plan is a sampling plan with an acceptance number of zero: one defective unit in the sample rejects the whole lot. It is sized from a rejectable quality level and consumer's risk (or AQL-indexed c=0 tables) and used where accepting a known defect is hard to defend.

What a c=0 plan is, and how one is built

Minitab describes c=0 plans, also called "accept on none" plans, as the ones used by organizations "required to reject an entire lot when one defect is found, such as with a zero-defectives program." Its help page gives two conditions for choosing one: "You have a specified or implied RQL and beta" and "You do not use an AQL." The RQL (rejectable quality level, often called LTPD) is the defect rate the buyer wants the plan to catch. Beta, the consumer's risk, is the chance the buyer is willing to accept a lot that bad anyway.

Minitab is plain about the price of that protection. If incoming quality is very good, a c=0 plan "can help you to minimize inspection quantities. However, if your incoming quality is not close to perfect, you will reject more lots and will likely inspect more total items." In other words, the plan suits a supplier whose process already runs close to zero defects, and punishes one that does not.

There are two common ways to get a plan. The first is published tables. CASRAI names Nicholas Squeglia's Zero Acceptance Number Sampling Plans as the most widely referenced tables for this approach. Mark Durivage, writing in Pharmaceutical Online, says Squeglia derived his plans from ANSI/ASQ Z1.4 but left out Z1.4's normal, tightened and reduced inspection, its switching rules and its double sampling, and that all you need to read the table is the lot size and a risk-based AQL. His worked example: a lot of 250 at AQL 1.5 calls for 19 pieces inspected with zero defects found. We do not reproduce the tables here; buy the book if a plan is going into a quality agreement. The second way is to compute the plan directly from an RQL and a beta, which is what Minitab does and what the next section shows.

The arithmetic behind a c=0 plan

With c=0, a lot passes only if every sampled unit is good. If a fraction p of a large lot is defective and units are drawn at random, the chance of that is (1 − p) multiplied by itself n times, where n is the sample size. Set that chance equal to beta at the RQL and solve for n:

n = ln(beta) / ln(1 − RQL), rounded up to the next whole unit.

Worked example (ours): a buyer wants no more than a 10% chance of accepting a lot that is 10% defective. n = ln(0.10) / ln(0.90) = −2.3026 / −0.1054 = 21.85, so the sample is 22 units. Check it the other way: 0.90 raised to the 22nd power is 0.098, just under the 10% target. The same formula gives these sample sizes for other targets:

RQL (lot % defective to catch)Beta (chance of accepting it anyway)n before roundingSample, c=0
10%10%21.922
10%5%28.429
5%10%44.945
1%10%229.1230
1%5%298.1299

These are our binomial calculations, not values copied from Squeglia or any standard. Two cautions apply. First, the formula assumes the sample is a small part of a large lot. For a small, isolated lot, Minitab's steps tick "Use hypergeometric distribution for isolated lot" and take an optional lot size; a reader reply on the Quality Digest article gives the rule of thumb that the hypergeometric is more precise once the sample exceeds about 5% of the lot. Second, the arithmetic only holds for a random sample. Durivage stresses that each piece "must have the same chance of being selected." Ten parts pulled from the top of one carton are not a random sample of the lot.

c=0 versus a Z1.4 AQL plan

c=0 acceptance sampling decision versus a traditional c greater than 0 planA sample of size n is pulled from the lot and inspected; under a c=0 plan any defective unit rejects the whole lot while zero defects accepts it, whereas a traditional c greater than 0 plan still accepts the lot when the defect count d is at or below the acceptance number c.No, zero defectsYes, one or mored is c or fewerd is more than c

Lot of parts to judge

Pull a random sample of size n and inspect it

c=0 plan: any defective unit in the sample?

ACCEPT the whole lot

REJECT the whole lot

Traditional plan: defects d vs accept number c, c over 0

ACCEPT, a few defects allowed

REJECT the lot

How a c=0 plan decides a lot versus a traditional c>0 plan. The same random sample of n pieces is inspected; under a c=0 plan any defective unit rejects the whole lot, while a traditional plan with an acceptance number c above zero still accepts the lot as long as the defect count d stays at or below c.
c=0 versus a traditional c>0 acceptance sampling plan, at a glance
Featurec=0 planTraditional (c>0) plan
Acceptance number (c)Zero. One defective unit in the sample rejects the whole lot.Above zero. The lot is accepted when the defect count d is at or below c.
Sample size (relative)Can be smaller when incoming quality is near perfect, but larger if matched to a traditional plan at a bad quality level; which is smaller depends on the point matched (illustrative — see the worked math above).Set from the AQL, lot size and inspection level; often a larger fixed sample that tolerates a stated number of defects.
Consumer (buyer) protectionStronger against good-but-imperfect lots: no sampled defective unit is ever accepted.Can accept a lot that holds confirmed defective units in the sample.
When to useZero-defect programs and safety-critical incoming inspection, such as many medical device and pharmaceutical buyers (CASRAI); often applied to the critical-defect class only.Most general US incoming-inspection plans under ANSI/ASQ Z1.4 / ISO 2859-1, with switching rules between lots.
FeatureANSI/ASQ Z1.4 AQL planc=0 plan
Plan indexed byAQL, lot size, inspection levelRejectable quality level (RQL/LTPD) and consumer's risk (Minitab); lot size and an AQL index in Squeglia-style tables (Durivage)
Acceptance numberOften above 0. QIMA's example: code letter L at AQL 2.5 is a sample of 200, accept on 10 or fewer defectives, reject on 11Always 0
Whose point of viewDesigned for roughly a 95% chance of accepting lots at the AQL (Quality Digest), with switching rules between lotsThe consumer's. A reader comment on the Quality Digest article says a plan that finds nonconformities "is doing its intended job"
Lot with a confirmed defective in the sampleCan still be acceptedAlways rejected
Typical userMost US incoming-inspection plans (CASRAI)Many medical device and pharmaceutical incoming programs, instead of or alongside Z1.4 (CASRAI)

A computed example: the same lots through both plans

Take the 22-piece c=0 plan from above and the Z1.4 plan QIMA quotes for code letter L at AQL 2.5 (200 pieces, accept on 10). The c=0 plan gives about a 10% chance of accepting a lot that is 10% defective. At 2.5% defective, which an AQL 2.5 program treats as acceptable, it accepts only about 57% of lots. The Z1.4 plan accepts about 99% of 2.5% lots and about 24% of 6.5% lots. That is the trade: c=0 protects the buyer sharply but rejects more lots from a supplier running near the AQL.

Operating characteristic curves: a c=0 plan with 22 samples falls steeply from the first defective percent, while the Z1.4 plan with 200 samples and Ac=10 stays near 100 percent acceptance up to about 3 percent defective and then drops sharplyTwo operating-characteristic curves compare how likely each plan is to accept a lot as its defect rate rises: the c=0 plan starts dropping the moment any defects appear, while the traditional Z1.4 plan tolerates a few percent defective before its acceptance probability falls off. Chance a lot is accepted vs its true percent defective (computed, binomial) 0%25%50%75%100%0%2.5%5%7.5%10%12.5%15% AQL 2.5 10% (RQL) Z1.4: n=200, Ac=10 c=0: n=22 true percent defective in the lot
True % defective in the lotc=0, n=22: chance of acceptanceZ1.4, n=200, Ac=10: chance of acceptance
1%80%over 99.9%
2.5%57%99%
4%41%82%
6.5%23%24%
10%9.8%0.8%

The curves cross between 6.5% and 7% defective. Above that point the 200-piece plan is the stricter one: at 10% defective it accepts under 1% of lots, against about 10% for the 22-piece c=0 plan. So "c=0" does not mean "stricter everywhere." It means stricter against good-but-imperfect lots and smaller per lot in this example, with less discrimination against very bad ones unless the sample is made larger. For how to read an operating characteristic curve in general, see the AQL sampling note.

Why the sources disagree on sample size

Here the published sources split. Quality Digest (William Levinson) says a c=0 plan "requires far less inspection than the corresponding ANSI/ASQ Z1.4 (formerly MIL-STD 105) plan, and becomes viable when the supplier is extremely confident in its level of quality." CASRAI says the "tradeoff is a larger required sample size to reach an equivalent confidence level." Which comparison holds for a given plan is not settled; it depends which two plans you match and on what point.

Our arithmetic shows why both statements can be true. The 200-piece Z1.4 plan above accepts a lot that is about 7.6% defective only 10% of the time. A c=0 plan with the same protection at that point needs ln(0.10) / ln(0.924) = 29.1, so 30 pieces: far fewer than 200, which is the Quality Digest direction. But that 30-piece plan accepts only about 47% of lots that are 2.5% defective, where the Z1.4 plan accepts about 99%. Ask a c=0 plan to accept 95% of 2.5% lots instead, and the formula allows a sample of only 2 pieces, which catches almost nothing. A single threshold of zero cannot hold both points of the curve at once, so which plan looks smaller depends on which risk you hold equal. This is our reading of the math, not a finding from either source.

Green molded plastic crate releasing from an open injection mold between blue and steel mold plates
A molded part leaving the mold. Under a c=0 plan, one confirmed defective unit in the sample, such as a crack or a short shot, is enough to reject the lot.

When buyers ask for c=0

CASRAI says many medical device and pharmaceutical incoming-inspection programs use zero acceptance number plans "instead of, or alongside, a Z1.4-derived AQL plan," because an AQL plan can accept a lot with confirmed defective units, which is "harder to defend for anything safety-critical." Durivage uses c=0 plans to size process-validation samples and writes that they "provide more protection to the consumer, which is especially important when health and human welfare are involved."

Outside regulated work the pattern is usually narrower. QIMA notes that because critical defects are unacceptable, "AQL for critical defects is commonly 0," while major and minor defects get ordinary AQL plans. Be clear about what that gives you: a zero-acceptance rule applied to an AQL-sized sample, not a plan sized from an RQL. Its protection is whatever that sample size delivers, which you can check with the same (1 − p) to the power n arithmetic. Nothing on this page claims that JCH or any partner factory holds a certification or works under these regulations; it describes how buyers write plans.

Switching rules belong to the AQL side

Z1.4 and ISO 2859-1 plans move between normal, tightened and reduced inspection based on lot history. Durivage notes that Squeglia's c=0 procedure did not integrate those rules, so if a quality agreement mixes the two systems, write down which rules apply to which defect class. On the AQL side, the public-domain MIL-STD-105E (section 4.8) and ISO 2859-1:1989 (clause 9.4) both say the acceptance procedures are discontinued when the cumulative number of lots not accepted on original tightened inspection reaches 5. Inspection resumes only after corrective action, and then on tightened inspection. That is 5 non-accepted lots counted since tightened inspection began, not necessarily 5 in a row.

Do not confuse this with the 10 accepted lots that MIL-STD-105E section 4.7.3 requires before reduced inspection. The exact wording in ANSI/ASQ Z1.4-2003 and in the second (1999) and third editions of ISO 2859-1 is not reproduced here; check the edition named in your contract. Quality Magazine calls Z1.4-2003 (R2018) "the current active standard" and reports that the third edition of ISO 2859-1 (final draft approved December 2025) adds skip-lot inspection to the switching rules.

What to settle before c=0 goes into a quality agreement

Common questions

Is a c=0 plan a replacement for ANSI/ASQ Z1.4?

No. William Levinson writes in Quality Digest that Z1.4 plans are designed to give "(roughly) a 95-percent chance of acceptance at the acceptable quality level", while Minitab says a c=0 plan is used when "You do not use an AQL." They answer different questions. CASRAI notes that medical device and pharmaceutical programs use c=0 "instead of, or alongside," a Z1.4-derived plan.

Why do medical device and pharmaceutical buyers use c=0?

CASRAI says an AQL plan can accept a lot that contains confirmed defective units, which is "harder to defend for anything safety-critical." Durivage, in Pharmaceutical Online, says c=0 plans give the consumer more protection, "especially important when health and human welfare are involved." This describes buyer practice; it is not a claim that JCH or any partner factory holds a certification or works to these regulations.

Does a c=0 plan mean a smaller sample?

Not settled as a general rule. Quality Digest says a c=0 plan "requires far less inspection than the corresponding ANSI/ASQ Z1.4" plan; CASRAI says the tradeoff is "a larger required sample size to reach an equivalent confidence level." Our own arithmetic shows either can come out true depending on which point of the curve the two plans are matched on. Quote both sources in any contract discussion.

Can I build a c=0 plan in Minitab?

Yes. Minitab's help page says the AQL is not required for a c=0 plan, but the software needs a value to run, so it suggests a very small one such as 0.05. It also says to use the highest alpha possible within alpha plus beta below 1; its example uses 0.89.

Does lot size change a c=0 plan?

For a large lot, hardly: the binomial formula uses only the sample size and the defect rate. For a small, isolated lot, Minitab's steps tick "Use hypergeometric distribution for isolated lot" and take an optional lot size. Table-based plans do move with lot size: Durivage's example needs 19 pieces for a lot of 250 at AQL 1.5 and two more for a lot of 500.

Where do I start if I need a normal Z1.4 plan?

Use the JCH AQL calculator to get the sample size and the accept and reject numbers for your lot size, inspection level and AQL. Then decide, defect class by defect class, whether anything critical should be held at zero accepted or handled with a separately sized c=0 plan.

Sources

Writing an inspection plan for a molded or assembled part? Send the drawing and your defect list, and we'll go through the defect classes and sampling approach with you before production starts. NDA first if you want one.

Send a part for reviewBook a 30-minute call

Ask JCH about c=0 sampling plans

Have a question about c=0 sampling plans? Ask our AI assistant. It answers from this website, and our tooling team follows up if you want.

AI assistant. Answers can be wrong; nothing it says is a quote or a promise.