Quality & InspectionFree referencePDF, Excel, Word exportDiscrete Manufacturing

Gage R&R Acceptance Criteria Reference Table

This table collects the acceptance thresholds quality engineers keep looking up when they review a gage R&R study: %GRR bands, ndc, resolution, bias, linearity, kappa and the calibration ratios. Search it, then export the rows you need into a procedure or supplier review.

Download:

Metric	Band or condition	Threshold	How it is usually read	Applies to
%GRR (of total variation)	Below 10%	< 10%	Measurement system generally acceptable	Variable data, AIAG MSA guidance
%GRR (of total variation)	10% to 30%	10 - 30%	May be acceptable depending on application, cost of gage, cost of rework or risk	Variable data, AIAG MSA guidance
%GRR (of total variation)	Above 30%	> 30%	Generally unacceptable, improve the system	Variable data, AIAG MSA guidance
%GRR (of tolerance)	Same bands as above	10 / 30%	Use when the gage is used for pass/fail against a specification	Variable data, product acceptance
%EV (repeatability)	Compared with %AV	No fixed limit	EV much larger than AV points to the gage or the method, not the operators	Variable data
%AV (reproducibility)	Compared with %EV	No fixed limit	AV much larger than EV points to training, fixturing or procedure	Variable data
ndc (number of distinct categories)	5 or more	>= 5	System can distinguish enough levels for process control	Variable data, AIAG MSA guidance
ndc (number of distinct categories)	2 to 4	2 - 4	Only coarse screening, not suited to process control charts	Variable data
ndc (number of distinct categories)	1	1	System cannot discriminate between parts	Variable data
Resolution (10:1 rule)	Gage increment vs tolerance or process spread	Increment <= 1/10	Smallest readable increment should be about one tenth of the tolerance or process variation	Gage selection
Test accuracy ratio (TAR)	Tolerance of UUT over calibrator accuracy	4:1 common target	Classic military-era rule, often still written into procedures	Calibration
Test uncertainty ratio (TUR)	Tolerance of UUT over expanded uncertainty of the calibration	4:1 common target	ANSI/NCSL Z540.3 uses 4:1 TUR as a fallback when probability of false accept cannot be estimated	Calibration
Bias	Observed average minus reference value	Not significantly different from 0	Judged with a confidence interval from the bias study, not a single reading	Variable data
Linearity	Bias across the operating range	Slope not significant	A significant slope means bias changes with the size of the part	Variable data
Stability	Control chart of a master over time	In statistical control	Out-of-control points signal drift, wear or environmental change	Variable data
Cohen kappa (attribute)	0.90 or higher	>= 0.90	Good agreement, appraisers accepted	Attribute data, AIAG MSA guidance
Cohen kappa (attribute)	0.75 to 0.90	0.75 - 0.90	Marginal, acceptable only with justification	Attribute data, AIAG MSA guidance
Cohen kappa (attribute)	Below 0.75	< 0.75	Poor agreement, improve criteria and training	Attribute data, AIAG MSA guidance
Effectiveness (attribute)	Correct decisions over opportunities	>= 90% acceptable, >= 80% marginal	Appraiser correct decision rate against the reference standard	Attribute data
Miss rate and false alarm rate	Bad called good, good called bad	Miss <= 2%, false alarm <= 5% (acceptable)	Customer-facing escapes weigh more than false rejects	Attribute data

These are commonly used guideline values from the AIAG MSA reference manual and ANSI/NCSL Z540.3 practice. Your customer, quality agreement or internal procedure may set stricter limits. Always confirm against the edition your customer requires.

Review our measurement system data model

Optional: a short review of how gage studies and calibration status could be tracked against your ERP item and equipment records.

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

How to use the table

Type a metric name such as ndc or kappa into the search box to filter the rows. Each row gives the band, the numeric threshold, how the result is normally read and which type of data it applies to. Use the variable data rows after an average and range or ANOVA study, and the attribute rows after an agreement study. Export to Excel when you want to paste the thresholds into a measurement system procedure.

Percent of what? Total variation versus tolerance

The same 10 and 30 percent bands are applied to two different denominators. Against total variation, %GRR tells you whether the gage can support process control and capability studies. Against tolerance, it tells you whether the gage can sort good from bad parts. A process with tight variation but a wide tolerance can look poor on the first and fine on the second. Report both and state which one your customer asks for.

Use total variation for SPC and capability work
Use tolerance for final inspection and product acceptance
Never mix the two when comparing studies

Why the 10 to 30 percent band needs a decision

Results between 10 and 30 percent are not an automatic pass. The guidance asks you to weigh the importance of the measurement, the cost of the gage, the cost of repairs and the risk to the customer. Record that reasoning and get customer agreement where the contract requires it. Results above 30 percent normally trigger a corrective action: better fixturing, higher resolution, clearer method or operator training.

Common mistakes

The most frequent errors are using parts that do not span the process range, which inflates the apparent gage error share, running too few parts, and treating a passing %GRR as proof the gage is calibrated. A gage R&R study measures variation, not accuracy against a standard. Pair it with calibration records and a bias study.

Frequently Asked Questions

What is an acceptable gage R&R percentage?

Under AIAG MSA guidance, under 10 percent is generally acceptable, 10 to 30 percent may be acceptable depending on the application and cost, and over 30 percent is generally unacceptable. Your customer may require a stricter limit, so check the quality agreement.

What does ndc of 5 mean?

Number of distinct categories is how many groups of parts the measurement system can reliably tell apart within the process spread. A value of 5 or more is the usual minimum for process control. It is computed from part variation divided by gage R&R, multiplied by 1.41.

Is the 4:1 ratio a requirement?

It is a long-standing rule of thumb, not a universal requirement. Many procedures still adopt 4:1 for test accuracy or test uncertainty ratio. Standards such as ANSI/NCSL Z540.3 define how decisions should be made when it cannot be met.

Do these limits apply to attribute gages?

Percent GRR bands apply to variable data. For attribute gages such as go/no-go or visual inspection, use kappa, effectiveness, miss rate and false alarm rate from an attribute agreement study instead.

If your gage R&R results live in a spreadsheet or a QMS, we can help pull them into your ERP quality module so acceptance status shows up on the gage record.

Talk to an Expert Explore More Free Tools

Related Tools

Quality & Inspection

More Quality & Inspection tools

See all

Back to All Tools