# ISO 13485 Sample Size: Confidence and Reliability Calculator

> Source: https://netray.co/tools/iso-13485-sample-size-calculator-aql-confidence-reliability
> Publisher: Netray (netray.co), AI and integration engineering for ERP and manufacturing systems.
> Section: Free Tools

## Short answer

Calculate a statistically justified sample size for ISO 13485 inspection and process validation using confidence and reliability, with a finite lot adjustment.

## Category

quality

## Industry

Medical Devices

## Audience

- Validation engineer
- Quality engineer
- Med device startup

## Tool

**Kind:** calculator

**Inputs:**

#### Confidence level

**Key:** conf

**Type:** select

**Default:** 0.95

**Options:**

##### 90%

**Value:** 0.9

##### 95%

**Value:** 0.95

##### 99%

**Value:** 0.99

**Help:** Probability that the claim about the population is true.

#### Reliability (proportion conforming)

**Key:** rel_pct

**Type:** slider

**Default:** 95

**Min:** 80

**Max:** 99.9

**Step:** 0.1

**Unit:** %

**Help:** Lowest conforming rate you want to demonstrate.

#### Lot or population size (0 if unlimited)

**Key:** lot

**Type:** number

**Default:** 0

**Min:** 0

**Step:** 50

**Unit:** units

**Help:** Used for the finite population adjustment.

#### Cost per unit tested

**Key:** unit_cost

**Type:** number

**Default:** 25

**Min:** 0

**Step:** 5

**Unit:** $

**Outputs:**

#### Sample size, zero failures allowed

**Key:** n0

**Expr:** ceil(ln(1 - conf) / ln(rel_pct / 100))

**Format:** integer

**Primary:** true

**Description:** Success run theorem: all samples must pass.

#### Adjusted for finite lot

**Key:** n_fin

**Expr:** if(gt(lot, 0), min(lot, ceil(n0 / (1 + (n0 - 1) / lot))), n0)

**Format:** integer

**Description:** Approximation. Never larger than the lot.

#### Maximum nonconforming rate demonstrated

**Key:** defect

**Expr:** 100 - rel_pct

**Format:** percent

#### Test cost

**Key:** cost

**Expr:** n_fin * unit_cost

**Format:** currency

**Chart:**

#### Sample size vs reliability at chosen confidence

**Type:** line

**X:** rel_pct

**From:** 80

**To:** 99.5

**Steps:** 40

**Series:**

- n0

**Disclaimer:** Zero-failure success run sizing for attribute (pass/fail) data. If you allow failures, variable data, or a regulator-specific plan, confirm the method with a statistician and your validation procedure.

## Content

**Intro:** ISO 13485 does not give you a fixed sample size. It expects you to justify one with a documented, statistically valid rationale for inspection and for process validation. This calculator gives the confidence and reliability success-run number you can cite and shows how lot size and reliability move it.

**Sections:**

#### How to use the calculator

**Body:** Pick the confidence level, set the reliability you must demonstrate and enter the lot size if the population is small. The primary output is the number of units that must all pass. Write the confidence, reliability and the reasoning in your validation protocol. Risk drives the choice: higher-risk characteristics under your risk management file deserve higher confidence and reliability.

**Bullets:**

- Link the confidence/reliability pair to the hazard severity in your risk file
- Document the formula and the rationale in the protocol, not just the number
- Re-run it when the claimed reliability changes

#### The formula and common pairs

**Body:** For pass/fail data with zero allowed failures, n = ln(1 - C) / ln(R), rounded up, where C is confidence and R is reliability. Common results: 90/90 needs 22 units, 95/90 needs 29, 95/95 needs 59, 99/95 needs 90 and 95/99 needs 299. If you allow one failure, the 95/90 plan rises to 46 and 95/95 to 93. More failures need a binomial calculation rather than this closed form.

#### Inspection sampling versus validation sampling

**Body:** Lot acceptance inspection often uses an AQL table such as ANSI/ASQ Z1.4 or ISO 2859-1, which sets a sample size from lot size and an acceptable quality level. That controls producer risk on a lot, but it does not by itself prove process capability. Process validation (IQ, OQ, PQ) commonly uses confidence and reliability for attribute data, or tolerance intervals and capability indices for variable data. Pick the method based on what the claim is.

#### Common mistakes

**Body:** Teams pick 30 units because it feels standard, with no link to risk. Others forget that the success run assumes independent, representative samples drawn across lots, operators and time, not a single batch built in one hour.

**Bullets:**

- Using one lot for a validation that claims ongoing process performance
- Applying the finite correction to justify a tiny sample on a small lot without a risk rationale
- Not documenting what happens if a unit fails

**Cta:** If your inspection and test records sit in an ERP or quality system, Netray can add a private AI layer on-premises so engineers can query historical results when writing a sampling rationale. Get in touch if useful.

**Faq:**

#### What sample size does ISO 13485 require?

**Answer:** The standard does not state a number. It requires that sampling plans be statistically valid and justified, and that process validation methods and criteria be documented. The justification usually links back to risk.

#### What does 95/95 mean?

**Answer:** It means 95 percent confidence that at least 95 percent of the population conforms. With zero allowed failures, that needs 59 samples.

#### Can I use AQL for process validation?

**Answer:** AQL plans control lot acceptance and are weak evidence of process capability. For validation, regulators and notified bodies generally expect a rationale tied to confidence, reliability or capability, plus risk.

#### Does a smaller lot let me use fewer samples?

**Answer:** A finite correction can reduce the number, and with very small lots you may test everything. Document the reasoning and make sure the samples still represent the real variation.

## Search terms

iso 13485 sample size, iso 13485 sample, iso 13485 sampling plan, medical device validation sample size, 95/95 sample size, success run theorem sample size, iso 13485 aql, sample size confidence reliability

---
Talk to an engineer: https://netray.co/contact?intent=erp_ai | Machine catalog: https://netray.co/agent/catalog.json