MSA Guide
Gage R&R Acceptance Criteria, Explained
The AIAG MSA manual gives you two numbers to pass: %GRR and ndc. Here's what the thresholds actually mean — and the decision most engineers get wrong in the 10–30% gray zone.
The AIAG thresholds
| %GRR | Verdict | What it means in practice |
|---|---|---|
| < 10% | Acceptable | Use the gage for any purpose, including capability studies and sorting. |
| 10–30% | Conditional | May be acceptable based on application importance, gage cost and rework cost. Document the justification. |
| > 30% | Unacceptable | The measurement system needs improvement before you trust any number it produces. |
| ndc ≥ 5 | Required | ndc = 1.41 × (PV/GRR). Below 5, the system can't distinguish parts well enough for SPC. |
Source: AIAG Measurement Systems Analysis (MSA) Reference Manual, 4th edition conventions.
Failing on repeatability vs reproducibility
- Repeatability (EV) dominates— the gage itself is the problem: worn fixture, loose clamping, insufficient resolution (the 10-to-1 rule), vibration. Retraining operators won't help.
- Reproducibility (AV) dominates — operators measure differently: inconsistent part loading, different reading technique, ambiguous work instructions. Fix the method, not the gage.
The sampling mistake that fakes a failure
%GRR = GRR/TV, and TV includes part variation. If your 10 study parts don't span real process variation — all pulled from one stable lot — PV shrinks, TV shrinks, and %GRR balloons. A good gage can fail the study because the parts were too similar. Select parts across the full process range, including near both spec limits.
Run a Gage R&R study in your browser
10 parts × 3 operators × 3 trials — paste your measurements, get %GRR, EV, AV, PV and ndc with verdicts. Free.
Open the Gage R&R calculator →FAQ
My %GRR is 25%. Can I still use the gage?
Conditionally, per AIAG — acceptable depending on the importance of the application, cost of the gage, and cost of rework. In practice: fine for monitoring a stable non-critical dimension, not fine for sorting parts against a tight spec or for capability studies you submit to customers.
What's the difference between %GRR of Total Variation and %GRR of Tolerance (P/T ratio)?
%GRR/TV compares measurement error to the process variation you observed; P/T compares it to the spec width. A gage can pass one and fail the other. Use P/T when the gage's job is checking conformance to spec; use %GRR/TV when it's monitoring process behavior. Report both when in doubt.
Why does ndc matter if my %GRR passed?
ndc (number of distinct categories) tells you how many groups of parts the measurement system can reliably distinguish. With ndc below 5, your control charts and capability studies are built on measurements that can't tell parts apart — even if %GRR looks acceptable on paper.
How many parts, operators and trials do I need?
The standard AIAG crossed study uses 10 parts × 3 operators × 3 trials (90 measurements). Parts must span the actual process variation — picking 10 parts from one lot made the same morning will inflate %GRR because PV shrinks.
Chapter 7 of the SPC Field Manual ($19) walks through a full worked Gage R&R with the variance math and a failing-study rescue plan.