The test uncertainty ratio (TUR) compares the tolerance being verified to the measurement uncertainty of the measurement used to verify it. For a two-sided tolerance of plus or minus T verified with an expanded uncertainty U at a coverage factor of k = 2, the TUR is simply T divided by U; expressed against the full specification span it is the tolerance span divided by twice the expanded uncertainty. The long-standing rule of thumb — originating in MIL-STD-45662A and carried into ANSI/NCSL Z540 — is that a TUR of 4:1 or better gives high confidence that the measurement can correctly resolve conformity. Below 4:1, the measurement uncertainty consumes too large a share of the tolerance, and the risk of a wrong accept/reject decision rises sharply unless guard banding is applied.
Take a characteristic with a tolerance of plus or minus 0.010 mm, verified by a calibration with an expanded uncertainty U of 0.002 mm (k = 2). The TUR is 0.010 divided by 0.002, or 5:1 — comfortably above the 4:1 threshold, so the measurement can resolve conformity with confidence. Now suppose the uncertainty is instead 0.004 mm: the TUR drops to 2.5:1, which is marginal. The measurement uncertainty is now 40% of the tolerance, and a reading near the limit could easily fall on the wrong side of the true value — exactly the situation where guard banding becomes necessary.
The 4:1 convention dates to mid-twentieth-century military calibration standards and reflects a practical balance: when measurement uncertainty is no more than about a quarter of the tolerance, the probability of falsely accepting a nonconforming item with simple acceptance stays low without forcing every measurement onto more expensive references. It is a rule of thumb, not a law of physics — modern practice under ISO/IEC 17025 emphasizes calculating the actual probability of false acceptance and applying an explicit decision rule rather than relying on a single ratio. But 4:1 remains a useful first screen for whether a measurement system is adequate for a tolerance.
TUR is often confused with the test accuracy ratio (TAR), but they differ. TAR compares the tolerance to the accuracy or specification of the reference standard alone, ignoring the full uncertainty budget — repeatability, environment, resolution, and so on. TUR uses the complete expanded measurement uncertainty and is therefore the more rigorous and more honest figure, because a reference with a great specification can still yield a poor TUR once all uncertainty contributors are included. When a requirement says 4:1, confirm whether it means TUR or TAR; modern standards favor TUR.
TUR and decision rules are two sides of the same problem. When the TUR is 4:1 or better, simple acceptance — accepting any in-tolerance result — carries acceptably low false-accept risk and is often allowed by agreement. When the TUR falls below 4:1, the measurement uncertainty is large relative to the tolerance, and a guard band should be applied so the acceptance limit moves inward and the probability of false acceptance stays controlled, consistent with ISO/IEC 17025 Clause 7.1.3 and ILAC G8. In other words, a low TUR does not forbid using the measurement — it triggers the need for an explicit decision rule.
If a measurement has a TUR below 4:1, you have three defensible options. Improve the measurement: use a lower-uncertainty reference standard or a better method to shrink U. Apply guard banding: keep the gage but move the acceptance limits inward to control false-accept risk, documenting the decision rule. Or agree shared risk with the customer, explicitly accepting the higher risk in writing. What is not defensible is making pass/fail decisions on a low-TUR measurement with simple acceptance and no acknowledgment of the risk.
CalibrationOS derives the test uncertainty ratio for each calibration from the expanded uncertainty in its budget against the tolerance being verified, surfaces it on the record, and flags low-TUR measurements so they can be routed to a guard-band decision rule rather than silently accepted. This connects the uncertainty budget, the TUR, and the conformity decision in one defensible chain.
TUR is the ratio of the tolerance being verified to the expanded measurement uncertainty of the measurement used to verify it. For a tolerance of plus or minus T and expanded uncertainty U at k = 2, TUR = T / U. A TUR of 4:1 or better is the common adequacy threshold.
Divide the tolerance by the expanded measurement uncertainty: for a plus-or-minus T tolerance and expanded uncertainty U (k = 2), TUR = T / U. For example, a plus-or-minus 0.010 tolerance measured with U = 0.002 gives a TUR of 5:1.
The 4:1 rule, from MIL-STD-45662A and ANSI/NCSL Z540, holds that the measurement's expanded uncertainty should be no more than a quarter of the tolerance. At 4:1 or better, simple acceptance carries low false-accept risk; below it, guard banding is recommended.
TAR (test accuracy ratio) compares the tolerance to the reference standard's accuracy or specification alone. TUR (test uncertainty ratio) uses the full expanded measurement uncertainty, including repeatability, environment, and resolution, making it the more rigorous and modern figure.
Improve the measurement with a lower-uncertainty reference or method, apply guard banding to control false-accept risk while documenting the decision rule, or agree shared risk with the customer in writing. Do not make pass/fail decisions on a low-TUR measurement with simple acceptance and no risk acknowledgment.
This article is licensed CC BY-SA 4.0. Share, adapt, and reuse with attribution to calibrationos.com/learn/test-uncertainty-ratio-tur.
Get expert insights on compliance, uncertainty, and measurement best practices.
Start managing calibrations in minutes. Free plan with 25 assets — no credit card.
Get Started Free