Sampling Inspection: Sample Size, Levels and Limits

A lot of 4,000 turned parts arrives, and quality assurance has to decide within a few hours: does a sampling inspection cover it, or does the lot need a full inspection instead? Anyone answering that purely by gut feeling either inspects too much and loses time, or too little and risks defective parts reaching their own production line. A properly planned sampling inspection resolves exactly that dilemma.

In brief: Sampling inspection under ISO 2859-1 sets the sample size from lot size and inspection level, and determines how many defects in that sample still allow the whole lot to be accepted. For safety-critical parts, this statistical safeguard alone is not enough.

What is sampling inspection in incoming goods control?

In a sampling inspection, only a statistically defined portion of a lot gets checked. The result from that subset decides whether the entire lot is accepted or rejected. How this fits into the broader incoming goods inspection process, and when a full inspection makes more sense instead, is covered in our foundational post on that topic.

The statistical basis for sampling inspection is DIN ISO 2859-1 ("Sampling procedures for inspection by attributes"). It defines how many parts must be drawn from a lot and how many defective parts are still tolerated without rejecting the whole lot.

We cover the actual AQL value tables and worked examples in detail in a dedicated post on Acceptable Quality Level (AQL). This post picks up one step earlier: how do you actually arrive at the right sample size, and what practical decisions sit in between?

How do you calculate sample size under ISO 2859-1?

Sample size for a sampling inspection comes from two inputs: lot size and the chosen inspection level. Together, these map to a sample size code letter, which in turn determines the actual sample size.

The table below shows an excerpt of common lot-size ranges for the frequently used Inspection Level II:

Example: A lot of 1,000 parts is inspected under Level II. That gives code letter J and a sample size of 80 parts. At an AQL of 1.5, up to 3 of those parts may be defective and the lot still gets accepted; at 4 or more defective parts, it's rejected. Our linked AQL post explains how to pick the right AQL value for different part types.

For a larger lot of 4,000 parts, the sampling inspection sample size only rises to 200 (code letter L), even though the lot size has quadrupled. That's the central advantage of sampling inspection: inspection effort grows far more slowly than lot size.

What do inspection levels I, II, and III mean?

In a sampling inspection, the inspection level determines how well a sample can discriminate between good and bad lots. Level II is the default and applies unless otherwise agreed.

• Level I: Smaller sample size, lower discriminating power. Defensible for suppliers with a demonstrated, stable process, not meant as a general cost-cutting shortcut.

• Level II: The standard case for most industrial goods.

• Level III: Larger sample size, higher discriminating power. Useful after quality problems occur, or for higher-risk parts.

• Special levels S-1 to S-4: Much smaller samples for cases where large samples are impractical, such as destructive testing. Explicitly unsuitable for safety-critical characteristics due to their lower discriminating power.

What changes between normal, tightened, and reduced inspection?

ISO 2859-1 defines three inspection severities that a sampling inspection switches between based on delivery history: normal, tightened, and reduced. This mechanism rewards reliable suppliers with less inspection effort and responds to quality problems with stricter criteria.

Switching to reduced inspection needs all three conditions at once, not just a lot count.

In our sourcing projects with Far East partners, we tie this switch closely to the ongoing supplier evaluation: a supplier delivering cleanly across multiple lots earns reduced inspection because the data supports it.

How reliable is a sampling inspection, really?

A sampling inspection gives a statistically grounded, but not an absolute, statement about a lot's quality. Even applied carefully, some residual risk remains that defective parts go undetected.

At an AQL of 1% and a sample size of 200, an accepted lot can still contain up to 4.59% defective parts without the sampling inspection necessarily catching it.

Part of the reason is that real-world defects often cluster in specific sections of a lot, for example when a production problem only starts appearing toward the end of a manufacturing run. A sampling inspection covering only the start of the lot can easily miss that kind of defect cluster.

Full inspection isn't a guarantee of complete defect detection either. Manual visual inspection realistically doesn't reach full detection rates in practice, fatigue and repetition effects limit even a 100% inspection. The choice between sampling inspection and full inspection is therefore mainly a question of the residual risk you're willing to accept relative to inspection effort.

When isn't a sampling inspection enough?

For safety-critical parts, relying on a sampling inspection alone is generally not acceptable given its statistical residual risk, and full inspection is standard practice. This applies in particular to brake system and airbag assemblies in the automotive industry and medical device components, as well as sensors and control components in aerospace, where standards such as AS9100 also mandate full inspection instead of pure sampling for critical assemblies. Full inspection also applies as a transitional measure whenever a manufacturing process hasn't yet demonstrated stable process capability.

With imported mechanical parts, we've found a simple rule of thumb in practice: whenever the cost of a potential defect reaching your own production or the end customer clearly outweighs the extra cost of full inspection, full inspection is almost always worth it, regardless of whether a sampling inspection would be statistically defensible.

For imported products, traceability adds another layer: anyone who needs to identify affected batches in case of a defect needs airtight documentation anyway, the kind described in our post on product safety in imports.

The table below summarizes when a sampling inspection is enough and when full inspection is the right call:

How can you further de-risk sampling for Far East shipments?

With Far East shipments, geographic distance amplifies the risk of an undetected defective batch. By the time a defect surfaces during incoming goods control in Germany, the goods have already spent weeks at sea, and a replacement shipment takes just as long. An inspection carried out directly at the manufacturer, before shipping, reduces this risk before freight and time costs are even incurred.

At Line Up, that's built into our sourcing process. Our quality assurance team on the ground in China applies the same system of inspection level, sample size, and switching rules before goods ship, not just at the German incoming-goods stage. For our customers, that means an added layer of protection without increasing inspection effort at their own site.

Frequently Asked Questions

How do you calculate sample size for incoming goods control?

Sample size comes from lot size and inspection level under ISO 2859-1. Together, these two values produce a sample size code letter, which determines the actual sample size.

What do the inspection levels under ISO 2859-1 mean?

Levels I, II, and III differ in how well the sampling inspection discriminates good from bad lots. Level II is the standard case, Level III is used for higher risk or after quality problems, and Level I applies to suppliers with a proven, stable track record.

When does inspection need to be full rather than sampling-based?

For safety-critical parts, such as in automotive, aerospace, or medical technology, and for manufacturing processes without demonstrated process capability, full inspection is standard practice instead of a sampling inspection.

What's the difference between normal, tightened, and reduced inspection?

The severity of a sampling inspection switches based on delivery history: repeated rejected lots trigger tightened inspection, while several cleanly accepted lots can, under specific conditions, allow a switch to reduced inspection.

How reliable is a sampling inspection, really?

A sampling inspection statistically reduces the risk of defective goods but doesn't eliminate it entirely. Clustered defects within a lot in particular can be missed by a sample.

Conclusion: Sampling inspection as a practical tool for incoming goods control

Sampling inspection under ISO 2859-1 makes incoming goods control economically viable at large lot sizes, but it doesn't replace a risk assessment. Lot size, inspection level, and switching rules provide a statistically sound tool, yet the decision for or against full inspection remains a question of actual risk and supplier track record.

At Line Up, we combine this system with upstream quality assurance directly at the manufacturer in the Far East, so your incoming goods control in Germany becomes an added confirmation step. 👉 Schedule a free consultation and let's work out together how to de-risk your incoming goods control for Far East shipments.