Measurement uncertainty: GUM, CISPR 16-4-2, decision rules

Hugues Orgitello updated 7 Oct 2026 15 min read

A borderline EMC or radio result is decided by three texts read together: the GUM, which says how an uncertainty is built; the measurement standard, which may fix the uncertainty a laboratory is allowed (CISPR 16-4-2 for emissions) or simply state its own rule; and ISO/IEC 17025, which says when the uncertainty must be reported and how a pass or fail statement must name its decision rule. This guide gives the actual numbers and clause references from those texts (U_cispr values, ILAC-G8 guard bands, the 17025 reporting clauses), the 2026 changes to the GUM suite, and the common readings that the texts contradict.

What changed, and when

Date	Document	What it does
2008	JCGM 100:2008 (GUM) and JCGM 101:2008 (Supplement 1, Monte Carlo)	Base method; Supplement 1 propagates distributions numerically
2011	JCGM 102:2011 (Supplement 2)	Extends the method to any number of output quantities
2012	JCGM 106:2012 and JCGM 200:2012 (VIM, third edition)	Uncertainty in conformity assessment (guard bands); vocabulary
June 8, 2011	CISPR 16-4-2 edition 2.0	U_cispr table and the U_lab/U_cispr compliance rule; FAR and telecom port methods added
2014, 2018	CISPR 16-4-2 amendments 1 and 2	Add further methods (CDNE, Δ-AN, LLAS among them); consolidated as edition 2.2. The IEC catalogue shows a revision under development
November 2016	ETSI EN 300 328 V2.1.1	Clause 5.2 lists maximum measurement uncertainties (Table 17) and compares the measured value directly with the limit
July 2019	ETSI EN 300 328 V2.2.2	Clause 5.2 becomes "Void"
September 2019	ILAC-G8:09/2019	Decision rules and guard bands for ISO/IEC 17025:2017
2020	JCGM GUM-6:2020, ILAC-P10:07/2020, ILAC-P14:09/2020	Measurement models; traceability policy (implementation July 2021); CMC and uncertainty on calibration certificates
January 2021	ILAC-G17:01/2021	Measurement uncertainty in testing
2022	ILAC-G24:2022 / OIML D 10:2022	Determination of recalibration intervals
April 2023	IEC Guide 115:2023, edition 3.0	Replaces the "accuracy method" with "simple acceptance" and adds decision rule content
2023	JCGM GUM-1:2023	Introduction to the GUM suite
January 1, 2026	Global ACI	Single MRA covering the scopes of the former ILAC MRA and IAF MLA
2026	JCGM 100:2008/Amd.1:2026 and JCGM GUM-5:2026	Amendment on non-linearity in measurement models (4.1.4 note, H.1.7 second-order terms); Part 5, worked examples

What ISO/IEC 17025:2017 actually requires

Clause	Requirement
3.7	Decision rule: "rule that describes how measurement uncertainty is accounted for when stating conformity with a specified requirement"
6.4, 6.5	Equipment calibration programme; metrological traceability to the SI through an unbroken chain of calibrations, each contributing to the measurement uncertainty
7.1.3	When the customer asks for a statement of conformity, the specification and the decision rule must be clearly defined and, unless inherent in the standard, agreed with the customer
7.6.1, 7.6.3	All significant contributions must be identified. A testing laboratory shall evaluate measurement uncertainty; where the method precludes a rigorous evaluation, an estimate based on principles or experience. Note 1: a well recognised test method that limits the major sources of uncertainty and fixes the form of results satisfies 7.6.3 when followed
7.7	Monitoring the validity of results, including comparison with other laboratories (proficiency testing, inter-laboratory comparisons) where available and appropriate
7.8.3.1 c	Test reports give the uncertainty where it is relevant to the validity or application of the results, where the customer requires it, or where it affects conformity to a specification limit
7.8.4.1 a	Calibration certificates always give the measurement uncertainty
7.8.4.3	No recommendation on the calibration interval on a certificate or label, unless agreed with the customer
7.8.6.1, 7.8.6.2	A documented decision rule, accounting for false accept and false reject risk; the statement identifies which results it covers, which specifications are met or not, and the decision rule applied (unless inherent in the standard)

Clause texts as quoted in ILAC-G8:09/2019, IEC Guide 115:2023 and the NIST SOP 1 checklist.

GUM framework, the essentials

Type A and Type B evaluation

Type A evaluation uses a series of repeated observations and statistical methods. Type B uses any other information: calibration certificates, manufacturer specifications, handbook data, earlier measurements, experience. GUM 3.3.4 states that the classification "is not meant to indicate that there is any difference in the nature of the components": both end up as standard deviations and are combined in the same way.

Standard uncertainty from a Type B half-width

Information available	Distribution	Standard uncertainty for half-width a	Text
Certificate gives U and k	Normal	U divided by k	GUM 4.3.3
Only bounds, no other knowledge	Rectangular	a divided by square root of 3	GUM 4.3.7
Values near the centre more likely than near the bounds	Triangular	a divided by square root of 6	GUM 4.3.9
Digital resolution δx	Rectangular of width δx	0.29 δx	GUM F.2.2.1
Mismatch at RF interfaces, cyclic effects	U-shaped (arcsine)	a divided by square root of 2	GUM H.1.3.4 (cyclic temperature example)

The spread between these divisors is real: for the same half-width, the U-shaped assumption gives a standard uncertainty √3 (about 1.73) times larger than the triangular one.

Combined and expanded uncertainty

The combined standard uncertainty u_c is the root sum of squares of the contributions c_i × u_i, plus covariance terms when inputs are correlated. CISPR 16-4-2 equation (1) uses the uncorrelated form. The expanded uncertainty is U = k × u_c. GUM 6.3.3: when the output is approximately normal with sufficient effective degrees of freedom, k = 2 gives an interval with a level of confidence of approximately 95%, and k = 3 approximately 99%. CISPR 16-4-2 equation (2) fixes U_lab = 2 u_c(y).

CISPR 16-4-2, the EMC emissions rule

CISPR 16-4-2 deals with measurement instrumentation uncertainty (MIU): the receiver, cables, transducers (AMN, clamp, antenna) and the test site. Uncertainties due to the EUT, its set-up and cable arrangement are not part of it; the introduction of the standard places those under CISPR 16-4-1. CISPR 16-4-2 is a basic EMC publication: whether a product standard applies its compliance criterion is decided by that product standard.

Table 1, values of U_cispr (edition 2.0, 2011)

Measurement	Frequency range	U_cispr (k=2)	Basis
Conducted emissions, mains port, AMN (LISN)	9 kHz to 150 kHz	3.8 dB	Table B.1
Conducted, mains port, AMN	150 kHz to 30 MHz	3.4 dB	Table B.2
Conducted, mains port, voltage probe	9 kHz to 30 MHz	2.9 dB	Table B.3
Conducted, telecommunication port, AAN	150 kHz to 30 MHz	5.0 dB	Table B.4
Conducted, telecommunication port, capacitive voltage probe	150 kHz to 30 MHz	3.9 dB	Table B.5
Conducted, telecommunication port, current probe	150 kHz to 30 MHz	2.9 dB	Table B.6
Disturbance power (absorbing clamp)	30 MHz to 300 MHz	4.5 dB	Table C.1
Radiated, OATS or SAC	30 MHz to 1000 MHz	6.3 dB	Tables D.1 to D.4
Radiated, FAR	30 MHz to 1000 MHz	5.3 dB	Tables D.5, D.6
Radiated, FAR	1 GHz to 6 GHz	5.2 dB	Table E.1
Radiated, FAR	6 GHz to 18 GHz	5.5 dB	Table E.2

Below 1 GHz the values were computed for the quasi-peak detector (average and r.m.s.-average assumed not to exceed them); above 1 GHz for the peak detector. Where the annexes give several values, U_cispr is the maximum. Amendments 1 and 2 add methods (CDNE, Δ-AN, LLAS) whose values are in the consolidated edition; check that edition for those rows. The first edition (2003) listed 4.0 dB and 3.6 dB for mains conducted, 4.5 dB for disturbance power and 5.2 dB for radiated 30 to 1000 MHz, which is where the figures still circulating come from.

Compliance assessment (clause 4.2)

Situation	Compliance is deemed to occur if
U_lab <= U_cispr	No measured disturbance level exceeds the limit
U_lab > U_cispr	No measured level, increased by (U_lab minus U_cispr), exceeds the limit

Clause 4.1 adds a reporting rule: if U_lab is at most U_cispr, the report may give U_lab or simply state that U_lab is less than U_cispr; if U_lab exceeds U_cispr, the report shall give the U_lab value for the instrumentation actually used. Nothing in the clause replaces the CISPR 16-1 equipment specifications or the CISPR 16-4-3 statistical requirements for series production.

What goes into a radiated U_lab

The input quantities common to all methods are the receiver reading, the attenuation of the connection between receiver and transducer, mismatch, and receiver corrections for sine wave voltage accuracy, pulse amplitude response, pulse repetition rate response and noise floor proximity. Radiated methods add antenna and site terms: antenna factor and its frequency interpolation, antenna factor variation with height, directivity, phase centre location, cross-polarisation, balance, site imperfections, separation distance and table effects. EUT positioning repeatability is not an MIU term: if a laboratory includes it, its U_lab is no longer comparable with U_cispr.

Radio: ETSI test standards follow their own rule

Radio conformance tests under ETSI harmonised standards do not use the CISPR excess rule. In EN 300 328 V2.1.1 (November 2016), clause 5.2 said the measured value is compared with the limit to decide conformity, the measurement uncertainty is included in the report, and each recorded uncertainty must be at or below Table 17 (for example ±1.5 dB for conducted RF output power, ±3 dB for conducted power spectral density and conducted unwanted emissions, ±6 dB for radiated emissions), computed at k = 1.96 or k = 2 using ETSI TR 100 028. In V2.2.2 (July 2019), clause 5.2 is "Void"; the annex on test sites still points to TR 100 028 parts 1 and 2 and TR 102 273 for computing site uncertainty. For a given radio report, read the version of the harmonised standard actually applied and its interpretation clause, if any. TR 100 028 itself is a technical report covering parameters such as power, modulation accuracy and spurious emissions; it is guidance, not a pass or fail rule.

ILAC-G8 decision rules

ILAC-G8:09/2019 takes JCGM 106:2012 as its reference. A guard band w is the distance between the tolerance limit TL and the acceptance limit AL.

Rule	Guard band w	Specific risk (ILAC-G8 Table 1)
Simple acceptance ("shared risk")	0	False accept probability below 50% (up to 50% for a result on the limit)
ISO 14253-1:2017	0.83 U	Below 5% false accept
ILAC G8:2009 rule	1 U	Below 2.5% false accept
"3 sigma"	1.5 U	Below 0.16% false accept
"6 sigma"	3 U	Below 1 ppm false accept
Uncritical (guarded rejection)	minus U	Reject only above TL + U: below 2.5% false reject
Customer defined	r × U	As agreed

Table 1 assumes a single-sided specification and a normal distribution. ILAC-G8 also describes a non-binary statement: pass below TL minus w, conditional pass between TL minus w and TL, conditional fail between TL and TL + w, fail above TL + w. It notes that the same result can pass with one guard band and fail with a larger one, which is why the rule must be agreed before measuring (17025 clause 7.1.3), and that where legislation mandates a rule, the laboratory follows it.

In EMC emissions testing the CISPR 16-4-2 criterion is the rule inherent in the method when the product standard applies it. A customer can still ask for a guard band on top for internal release decisions.

For electrical safety, IEC Guide 115:2023 (4.3.3) states that within the IECEE CB Scheme results are compared with the limits without applying measurement uncertainty ("simple acceptance"), with equipment accuracy meeting the test standard or IECEE OD-5014 defaults, and suggests a report statement citing Guide 115 4.3.3 to satisfy 17025 clause 7.8.6.1.

Mismatch, the term most often misread

For two ports with reflection coefficients Γ1 and Γ2, the mismatch term lies between 20 log10(1 minus |Γ1||Γ2|) and 20 log10(1 plus |Γ1||Γ2|) and is treated as U-shaped. Example: an antenna with VSWR 2 (|Γ| = 0.33) on a cable chain with VSWR 1.5 (|Γ| = 0.2): |Γ1||Γ2| = 0.067, bounds of about +0.56 dB and minus 0.60 dB, so a half-width of about 0.6 dB and a standard uncertainty of about 0.42 dB (half-width divided by √2). A 10 dB attenuator at the antenna port improves the return loss seen from the cable by 20 dB (|Γ| divided by 10), shrinking the term at the cost of 10 dB of sensitivity.

Traceability, CMC and certificates

Routes accepted by ILAC-P10:07/2020 (implementation July 2021) when traceability is required:

1. A National Metrology Institute whose service is covered by the CIPM MRA (check the BIPM KCDB, the authoritative source, rather than a logo).
2. A calibration laboratory accredited for that calibration by a body covered by the ILAC Arrangement or a recognised regional arrangement. Since January 1, 2026 these scopes are carried by the Global ACI MRA.
3. Only when 1 or 2 is not possible: an NMI service outside the CIPM MRA, or a laboratory outside those arrangements, under a policy set by the accreditation body.

ILAC-P10 also warns against the "persistent misconception" that traceability is "to a specific National Metrology Institute": traceability belongs to measurement results and reference values, not to an organisation.

CMC versus reported uncertainty. Per ILAC-P14:09/2020, a CMC is the capability "available to customers under normal conditions" as stated in an accredited scope or in the KCDB, expressed at about 95% coverage and including the contribution of a "best existing device". Certificates must state the coverage factor and coverage probability. Because the customer device replaces the best existing device, reported uncertainties "tend to be larger" than the CMC. Testing laboratories are not expected to evaluate a CMC.

Recalibration intervals. A calibration certificate shall not carry a recommended interval unless the customer agreed (17025 clause 7.8.4.3). The user laboratory sets intervals and reviews them; ILAC-G24:2022 lists the factors (required uncertainty, type of equipment, tendency to wear and drift, extent and severity of use, environment, intermediate check results, transport, legal requirements) and states that no single best practice exists.

Proficiency testing

Clause 7.7 asks the laboratory to monitor its performance by comparison with other laboratories where available and appropriate. Proficiency results are usually scored as z = (x minus X) / σ_pt, where X is the assigned value and σ_pt the standard deviation for proficiency assessment set by the scheme provider (not necessarily the spread of the participants). Under ISO 13528: |z| ≤ 2.0 satisfactory, 2.0 < |z| < 3.0 questionable (warning signal), |z| ≥ 3.0 unsatisfactory (action signal).

Worked example (hypothetical numbers)

EUT at 100 MHz, measured 38.0 dBuV/m quasi-peak at 3 m in a SAC. EN 55032 Class B radiated limit at 3 m, 30 to 230 MHz: 40 dBuV/m quasi-peak.

Laboratory situation	Rule applied	Value compared with 40 dBuV/m	Verdict
U_lab = 5.5 dB (at or below U_cispr = 6.3 dB)	CISPR 16-4-2, 4.2	38.0	Compliant
U_lab = 7.1 dB (above U_cispr)	CISPR 16-4-2, 4.2	38.0 + 0.8 = 38.8	Compliant
U_lab = 7.1 dB, customer guard band w = U (ILAC-G8)	Acceptance limit 40 minus 7.1 = 32.9	38.0	Not accepted under that rule
U_lab = 7.1 dB, non-binary statement with w = U	ILAC-G8 4.2.3	38.0, between 32.9 and 40	Conditional pass

The same measurement passes the regulatory rule and fails an internal guard band rule: this is the case ILAC-G8 has in mind when it says conformity "is inherently connected to the decision rule employed".

Reading a test report budget annex

1. Which rule applies: CISPR 16-4-2 clause 4.2 (emissions), the interpretation clause of the harmonised standard version applied (radio), or a contracted ILAC-G8 rule.
2. U_lab per method with k stated, and the U_cispr row it is compared with (SAC 30 to 1000 MHz is 6.3 dB, not 5.2 dB).
3. Contributor list limited to instrumentation for the CISPR comparison; EUT and set-up terms belong elsewhere.
4. Excess added when U_lab exceeds U_cispr.
5. Decision rule named on any statement of conformity (17025 clause 7.8.6.2), unless inherent in the standard.
6. Calibration status of receiver, antenna, AMN and cables at the test date, against the laboratory own interval policy.

For the radiated emission test whose results this guide qualifies, see radiated emissions. For the immunity side, see IEC 61000-4-3. For the test sites behind the site imperfection term, see EMC chambers, SAC, FAR, OATS, reverb. For setups without a formal budget, see pre-compliance, TEM cell, near-field.

Claims the texts contradict

Common claim (some made earlier on this page)	What the text says
"U_cispr is about 5.2 dB for radiated emissions 30 to 1000 MHz and 3.6 dB for conducted"	Those are first edition (2003) values. CISPR 16-4-2:2011 Table 1: 6.3 dB at an OATS or in a SAC, 5.3 dB in a FAR, 3.4 dB for mains conducted 150 kHz to 30 MHz, 4.5 dB for disturbance power
"Above 1 GHz U_cispr is about 5.7 dB"	5.2 dB from 1 to 6 GHz and 5.5 dB from 6 to 18 GHz, in a FAR (Table 1)
"EUT positioning and cable layout are among the largest terms of U_lab"	U_lab is measurement instrumentation uncertainty only; EUT and set-up uncertainties are outside CISPR 16-4-2
"A test report without a stated uncertainty is incomplete under 17025"	7.8.3.1 c requires it only when relevant to validity or application, when the customer requires it, or when it affects conformity to a limit; 7.6.3 Note 1 covers methods that fix the uncertainty
"ETSI radio standards use the same excess rule as CISPR"	EN 300 328 V2.1.1 compared the measured value directly with the limit and capped the uncertainty (Table 17); V2.2.2 made clause 5.2 void
"ILAC-G8 has four canonical rules including a Bayesian decision"	ILAC-G8 describes simple acceptance, binary acceptance with a guard band and a non-binary statement, with Table 1 guard band multiples; it refers to JCGM 106 for the mathematics
"Guard band w = U gives about 95%, 2U gives about 99%"	ILAC-G8 Table 1: w = U keeps the specific false accept risk below 2.5%, 1.5 U below 0.16%, 3 U below 1 ppm (single-sided limit, normal distribution)
"Every calibration certificate must show the next due date"	17025 clause 7.8.4.3: no interval recommendation on a certificate or label unless agreed with the customer
"A certificate outside the ILAC MRA is not traceable"	ILAC-P10 accepts CIPM MRA NMI services, and in restricted cases other metrology institutes or laboratories under an accreditation body policy
"Traceable to NIST (or PTB, NPL)" as a guarantee	ILAC-P10 calls the link between traceability and a specific NMI a "persistent misconception"
"Picking the wrong distribution changes u by at most 1.22"	Between U-shaped (√2) and triangular (√6) divisors the factor is √3, about 1.73
"The z score divides by the standard deviation of the participants"	It divides by σ_pt, the standard deviation for proficiency assessment chosen by the provider
"The ILAC MRA is the current worldwide arrangement"	Since January 1, 2026, the Global ACI MRA covers the former ILAC MRA and IAF MLA scopes; existing accreditations stay recognised during the transition

Going further

• Radiated emissions, EMC test: the level measurement whose uncertainty this guide covers
• IEC 61000-4-3, radiated RF immunity: the immunity counterpart, with its own field uniformity requirements
• EMC chambers, SAC, FAR, OATS, reverb: the test sites behind the SAC and FAR rows of Table 1
• Pre-compliance, TEM cell, near-field: setups where no formal budget applies
• Conformity assessment bodies and accreditation: Global ACI, ISO/IEC 17025 and who accredits whom
• Decibels for EMC: the logarithmic units in which U_lab and limits are compared
• Glossary: definitions of EMC, CISPR, IEC and related terms

See also

• PCB design for EMC: return paths, decoupling, stackup
• Antenna design and impedance matching for IoT
• Radio: RX blocking, selectivity and intermodulation tests
• Radiated emissions EMC test: pre-scan and final scan

Working on reviewing the uncertainty budget annex of an EMC or radio test report and supporting marginal pass/fail decisions for a real product? AESTECHNO can audit your design and certification plan. Get in touch

A 30-minute audit with AESTECHNO engineers.

AESTECHNO is the electronics design house that edits spilma. We help product teams scope their certification effort and avoid the common mistakes.

Book a free audit

Sources & references

1. JCGM 100:2008, Evaluation of measurement data, Guide to the expression of uncertainty in measurement (GUM), BIPM JCGM www.bipm.org/en/committees/jc/jcgm/publications
2. JCGM 100:2008/Amd.1:2026, Nonlinearity in measurement models, BIPM JCGM www.bipm.org/en/doi/10.59161/jcgm100-amd-2026
3. JCGM GUM-5:2026, Guide to the expression of uncertainty in measurement, Part 5: Examples, BIPM JCGM www.bipm.org/en/doi/10.59161/jcgmgum-5-2026
4. JCGM 106:2012, The role of measurement uncertainty in conformity assessment, BIPM JCGM www.bipm.org/documents/20126/2071204/JCGM_106_2012_E.pdf/fe9537d2-e7d7-e146-5abb-2649c3450b25
5. JCGM 200:2012, International Vocabulary of Metrology (VIM), BIPM JCGM www.bipm.org/en/committees/jc/jcgm/publications
6. CISPR 16-4-2:2011+AMD1:2014+AMD2:2018, Measurement instrumentation uncertainty (catalogue entry), IEC CISPR webstore.iec.ch/publication/54
7. CISPR 16-4-2:2011 publisher preview (scope, Table 1, clause 4.2), via iTeh reseller, IEC via iTeh (secondary host) cdn.standards.iteh.ai/samples/16836/3d649487728c4d77ae1b73d819da7aa5/CISPR-16-4-2-2011.pdf
8. Uncertainty in RF emission measurements: revision of CISPR 16-4-2 (M. Stecher, first edition values and MIU scope), secondary source, Politecnico di Milano EMC chapter (secondary) emc-chap-it.polimi.it/download/MUinEMC_Revision%20of%20CISPR%2016-4-2%20doc.pdf
9. ISO/IEC 17025:2017, General requirements for the competence of testing and calibration laboratories, ISO IEC www.iso.org/standard/66912.html
10. ILAC-G8:09/2019, Guidelines on Decision Rules and Statements of Conformity, ILAC ilac.org/publications-and-resources/ilac-guidance-series/
11. ILAC-P10:07/2020 and ILAC-P14:09/2020, traceability and measurement uncertainty in calibration, ILAC ilac.org/publications-and-resources/ilac-policy-series/
12. IEC Guide 115:2023, Application of measurement uncertainty to conformity assessment activities in the electrotechnical sector, IEC webstore.iec.ch/en/publication/79233
13. ETSI EN 300 328 V2.1.1 (2016-11), clause 5.2 and Table 17, ETSI www.etsi.org/deliver/etsi_en/300300_300399/300328/02.01.01_60/en_300328v020101p.pdf
14. ETSI EN 300 328 V2.2.2 (2019-07), ETSI www.etsi.org/deliver/etsi_en/300300_300399/300328/02.02.02_60/en_300328v020202p.pdf
15. ETSI TR 100 028, Uncertainties in the measurement of mobile radio equipment characteristics, ETSI www.etsi.org/standards-search?search=TR+100+028
16. ISO/IEC 17025:2017 section 7.8 certificate review checklist (SOP 1, Appendix B), NIST www.nist.gov/document/sop-1-calibration-certificate-eval-app-b-c-20190506pdf
17. Understanding PT performance assessment (z scores per ISO 13528), secondary source, Eurachem www.eurachem.org/images/stories/leaflets/pt/pt_perf_assmnt/PT_Understand_Evaluation_V1_EN.pdf
18. About the Global ACI MRA, Global ACI global-aci.org/en/about-the-global-aci-mra/
19. BIPM Key Comparison Database (KCDB), BIPM www.bipm.org/kcdb/