Quality & InspectionFree calculatorPDF, Excel, Word export

Zero-Failure Reliability Demonstration Sample Size Calculator

Validation protocols often say test enough units to show 95 percent reliability at 95 percent confidence. This calculator gives the number, and shows what it costs.

Your numbers

Reliability to demonstrate

Minimum probability that a unit meets the requirement.

Confidence level

Test duration per unit

hours

Cost per test unit

Your results

Units required (zero failures allowed)

Success run theorem: n = ln(1 - C) / ln(R).

Test unit cost

$70,800

Total unit test hours

29,500

Confidence actually achieved with this n

95.2%

Valid for pass/fail (binomial) data with zero failures. One failure requires a larger sample from the full binomial relationship.

Save results:

Required sample size vs reliability target

Sweeps Reliability to demonstrate from 80 to 99.5, holding your other inputs fixed.

Validation traceability

Optional review of lot and test traceability in your ERP for validation evidence.

Optional. The tool and every export work without this. No spam.

How to use the calculator

Enter the reliability you must demonstrate and the confidence level your protocol or customer requires. Add test duration and unit cost to see the total effort. The result is the number of units that must all pass; a single failure invalidates the demonstration and you would need to test more or fix the design. The chart shows how steeply the sample size climbs as the target approaches 100 percent.

The success run theorem

If a unit has reliability R, the probability that n independent units all pass is R to the power n. To claim at least R with confidence C, that probability must be no more than 1 minus C. Solving gives n = ln(1 - C) / ln(R). For 95 percent reliability at 95 percent confidence the answer is 59 units; for 90 at 90 it is 22. Round up, never down.

Common pairs: 90/90 needs 22, 95/95 needs 59, 99/95 needs 299.
The test must represent worst case conditions to support the claim.

Using it in risk-based validation

Many quality systems tie sample size to risk: higher severity failure modes get higher reliability and confidence targets. Link the choice to the severity in the FMEA and record the rationale in the validation plan. Where destructive testing is costly, consider stress testing or a model-based approach and justify the departure.

Common mistakes

Using this for variable data wastes samples, since a variable-data tolerance interval needs fewer units. Another error is drawing all units from one lot, which shows only lot-level reliability. Finally, do not pad the sample after seeing a failure; define the plan before testing.

Frequently Asked Questions

Why 59 units for 95/95?

ln(0.05) divided by ln(0.95) is about 58.4, so you must test 59 units with zero failures to claim 95 percent reliability at 95 percent confidence.

What if one unit fails?

The zero-failure plan fails. You need a larger plan that allows one failure, computed from the binomial distribution, or you must fix the cause and retest.

Is this the same as AQL sampling?

No. AQL plans control lot acceptance risk. This demonstrates a reliability claim with a confidence level, typically for validation.

Can I use this for software or services?

Yes for any independent pass/fail trials, such as transactions or cycles, provided each trial is representative and independent.

Netray can help assemble lot and test traceability from your ERP so validation samples are properly representative and documented.

Talk to an Expert Explore More Free Tools

Related Tools

Engineering & PLM

More Quality & Inspection tools

See all

Back to All Tools