Go/no-go gauges: how to size them, the 10% rule, and when not to use them

The fastest inspection instrument on any shop floor has no display, no batteries and no calibration drift mid-shift: a hardened plug with two ends. Here is how go/no-go gauging actually works — Taylor's principle, how gauge limits are calculated from the work tolerance, why the GO end wears out first, and the one thing attribute gauging can never tell you.

How go/no-go gauging works

A go/no-go gauge converts a measurement into a decision. Instead of reading a value, the operator applies two fixed-limit ends:

• The GO end is made to the maximum material limit of the feature. It must enter (hole) or pass over (shaft) every conforming part.
• The NO-GO end is made to the least material limit. It must not enter or pass. If it does, the feature is oversize (hole) or undersize (shaft).

GO enters, NO-GO doesn't: accept. Anything else: reject. A trained operator checks a bore in 3–4 seconds, against 15–20 seconds with a bore gauge and no transcription errors. That is why high-volume machining lines still gauge 100% of critical bores decades after CMMs arrived.

Taylor's principle: why GO is full-form

William Taylor's 1905 rule still governs gauge design:

• The GO gauge checks size and form together — it should be a full-form gauge engaging the whole feature at once (a cylindrical plug the full length of the bore). A bore can be at perfect diameter at every cross-section and still trap a bent axis; a full-form GO plug catches it.
• The NO-GO gauge checks size only, ideally at individual cross-sections — which is why NO-GO ends are often shorter, or ball-ended, or a flat-anvil snap gauge. A lobed bore might rotate past a full-form NO-GO cylinder even when local diameters are out.

This is the same logic behind the envelope requirement in ISO 14405 and Rule #1 in ASME Y14.5: maximum material condition demands perfect form.

Gauge types

For thread gauges, remember the GO end verifies the whole thread — flank angle, lead and pitch diameter acting together — while the NO-GO checks only that the pitch diameter has not gone over the limit. Thread designations and pitch data are in the thread pitch calculator and the thread pitch reference.

Sizing a plug gauge: a ⌀25 H7 worked example

A ⌀25 H7 bore has limits 25.000 / 25.021 (tolerance 21 µm — check any fit with the ISO 286 fits calculator). The gauge maker cannot make gauges to zero tolerance either, so gauge standards (IS 3455 / ISO 1938 practice) allocate a gauge tolerance of roughly 10% of the work tolerance, here about 2 µm per end.

Both gauge tolerances point into the work tolerance band. The consequence is deliberate: the gauge can only reject a borderline-good part, never accept a bad one. You give up a sliver of your manufacturing tolerance (2–4% here) to guarantee that anything the gauge passes is truly in specification.

Wear allowance and calibration

The GO end rubs the full feature surface on every single part, so it wears toward accepting bad parts — a worn GO plug shrinks and starts entering undersize bores; a worn GO ring grows and starts passing oversize shafts. To buy life, a wear allowance (typically another ~5% of work tolerance, applied only when work tolerance is generous) is added on the GO end beyond the maximum material limit direction.

• Calibrate on a schedule driven by usage, not the calendar alone — a GO plug gauging 2,000 parts per shift needs checking monthly; a drawer gauge can run annually.
• Reject criterion: the GO end is scrapped when wear consumes the wear allowance; the NO-GO end almost never wears because it should rarely touch the work.
• Chrome-plated or carbide gauges extend GO life 5–10× in abrasive materials like cast iron — worth it above roughly 500 checks per day.

What attribute gauging can't tell you

A go/no-go decision contains no number. That has three hard consequences:

1. No process capability. You cannot compute Cp/Cpk from pass/fail data in any useful sample size — capability studies for PPAP need variable measurements. If the customer requires Cpk ≥ 1.33 on that bore, you need a bore gauge, air gauge or CMM for the study, even if production gauging is attribute.
2. No drift warning. The gauge tells you the process failed only after it has failed. SPC on measured values warns you while parts are still good.
3. Attribute agreement matters. Gauge R&R for attribute gauges exists (kappa studies) — two operators disagreeing on borderline parts is a real, measurable problem. The variable-gauge version is covered in our Gauge R&R guide.

The practical split most plants land on: variable measurement for setup approval, first-off and capability; attribute gauging for 100% in-process checks.

Frequently asked questions

Which end of a plug gauge is GO?

The longer end. GO is full-form and engages the whole bore length per Taylor's principle; NO-GO is deliberately short because it checks size at a cross-section. On double-ended gauges the NO-GO end also usually carries a red ring or groove.

What is the 10% rule for gauge tolerance?

The gauge maker's tolerance on each end is set at roughly 10% of the workpiece tolerance, placed inside the work tolerance band so the gauge can reject good parts but never accept bad ones. A 21 µm H7 bore gets about a 2 µm gauge tolerance per end.

Can I use go/no-go gauges for PPAP capability studies?

No. Capability indices (Cp, Cpk, Ppk) need variable data. Use a bore gauge, micrometer, air gauge or CMM for the study, then gauge 100% in production if you wish.

Why does only the GO end wear out?

The GO end passes through or over every conforming part, so it sees continuous rubbing contact. The NO-GO end should rarely enter anything — if your NO-GO is wearing, your process is producing parts at the wrong end of the tolerance.