Skip to main content
    QHSEtech
    Updated
    All glossary terms
    Content date: · No individual check logged· AI use

    Measurement System AnalysisMSA

    Also known as: Measurement systems analysis, Gauge R&R study, Gage R&R

    Definition

    Studies of how much a measuring process adds to the variation in its results: repeatability, reproducibility, stability, bias and linearity of gauges in use.

    Editorial definition: The QHSE Standard, updated 3 October 2026 · https://qhsetech.com/glossary/measurement-system-analysis

    Scope: MSA is not calibration: a calibrated gauge can still vary between operators or drift in use. Acceptance limits come from customers or from manuals such as AIAG's, which was not read.

    Example for your evaluation: Ask a tool to show a gauge study's raw readings, variance components and decision next to the gauge's calibration status and tolerance.

    Sources checked: . Examples are editorial illustrations.

    What NIST's handbook, the VIM, ISO, AIAG and IATF pages say about measurement system analysis

    On this page measurement system analysis, MSA, means studying the measuring process where it is used, with its gauges, operators, parts, methods and surroundings, to find how much of the variation in results comes from measuring itself. The NIST/SEMATECH e-Handbook of Statistical Methods calls the central study a gauge R&R study and analyses variability in three nested levels: level 1 is repeatability, the short-term precision of repeated readings; level 2 is reproducibility, which it treats as day-to-day variation; and level 3 is stability, the variation from run to run over a longer period. It combines them into the standard deviation behind the uncertainty of one measurement on a test item, and warns that other sources may need adding. This is general information, not certification advice.

    Other texts use reproducibility more widely. The VIM's reproducibility condition changes the location, the operators and the measuring system, and its note asks a specification to say which conditions were changed and which were not. A reproducibility figure from an operator study and a day-to-day figure from NIST's design are therefore different numbers, and a report should name the conditions varied.

    NIST follows the VIM in defining bias as the difference between a measurement result and its unknown true value, and says calibration to a reference standard can often estimate or remove it. It warns that calibration relates output to the true value in an ideal environment and may not show how a gauge behaves at work, where temperature, humidity, operators and wear add bias; the gauge study is meant to find those effects. Its list of sources of bias covers resolution, linearity, drift, hysteresis and differences among gauges, geometries and operators. For linearity it fits measured values against more than four reference standards spread over the range, each measured more than once: a slope other than one means the gauge is non-linear and needs calibration or repair, and an intercept other than zero means bias. For causes of non-linearity, such as poor calibration at the ends of the range and wear, it points to the list in the MSA reference manual.

    ISO's record shows ISO 22514-7:2026, on the capability of measurement processes, published in September 2026 as the third edition, replacing the 2021 edition and its 2024 amendment. Its abstract describes a procedure to judge whether a measuring system and process can meet a given task, with capability figures or ratios as acceptance criteria, following the GUM approach to combining uncertainty components. AIAG sells Measurement Systems Analysis, fourth edition, as manual MSA-4. The terms %GRR, commonly described as the gauge R&R spread as a percentage of total variation or of the tolerance, and ndc, the number of distinct categories of parts a measuring process can tell apart, are usually attributed to that manual and appear neither on its product page nor on the NIST pages read. IATF's 2026 news includes a communiqué of 30 July 2026 on the status of a second edition of IATF 16949, and a carmaker's production part approval requirements, effective 15 June 2026, that add a clarification on gauge R&R. The ISO, AIAG and IATF texts were not read.

    What we could not verify on 22 October 2026: how the AIAG manual defines and calculates %GRR and ndc, and any acceptance limits for them; the publication date of the fourth edition of the AIAG manual; the clause of IATF 16949:2016 that asks for measurement systems analysis, commonly cited as 7.1.5.1.1; and the content of the IATF communiqué, which is a PDF that was not opened.

    Sources opened :

    What a measurement system analysis record holds in software

    • The measuring process studied: gauge identity, characteristic, method, location and the tolerance it serves.
    • The study design: parts or check standards, operators, repeats, days or runs, and the order of measuring.
    • Every raw reading, tied to part, operator, trial and time, so that the analysis can be rerun.
    • The variance components found, such as repeatability, reproducibility, part-to-part variation and stability, with the calculation method named.
    • Bias and linearity results against reference values, with the fitted slope and intercept and their standard deviations.
    • The acceptance criteria applied, who set them, and the decision taken on the gauge.
    • The gauge's calibration status on the study date, and links to any action raised and to reports that used the gauge.

    MSA, calibration, gauge R&R, measurement uncertainty and process capability

    Calibration relates one instrument's indications to a reference under stated conditions, often in a laboratory. MSA looks at the measuring process where it is used, so a calibrated gauge can still fail a study. Gauge R&R is one MSA study, of repeatability and reproducibility; bias, linearity and stability studies are others. Measurement uncertainty describes the doubt attached to a result, and NIST's handbook turns its variability levels into the standard deviation behind it. Process capability compares the process spread with the specification, and the observed spread includes measurement variation.

    A hypothetical example: a gauge study before a capability report

    Everything in this case is invented and describes no real company. A plant making brackets must report capability on a hole position whose tolerance is 0.60 mm wide, from 1.70 mm to 2.30 mm. Its first report, from 50 parts, gives an observed standard deviation of 0.10 mm, so Cp is 0.60 / (6 x 0.10), which is 1.00. Before sending it, the quality engineer runs a gauge study with 10 parts, 3 operators and 2 repeats each, 60 readings in all, and finds a combined repeatability and reproducibility standard deviation of 0.06 mm.

    Assuming measurement and part variation are independent, they add as variances, so the part-to-part standard deviation is the square root of 0.10 squared minus 0.06 squared, which is 0.08 mm. The gauge alone is 0.06 / 0.10, or 60% of the observed standard deviation, and its spread of six standard deviations, 0.36 mm, is 60% of the tolerance. Without that spread, Cp would be 0.60 / (6 x 0.08), which is 1.25. The engineer does not report the higher figure. The gauge goes for repair and a new study, and the capability report waits for both, with the reasons recorded.

    Common misunderstandings about measurement system analysis

    • A calibrated gauge has not passed an MSA. A laboratory certificate describes the gauge under reference conditions, not with your operators, parts and workplace.
    • Reproducibility has more than one meaning. NIST's handbook uses it for day-to-day variation, while the VIM's reproducibility condition changes locations, operators and measuring systems.
    • Gauge R&R is not the whole of MSA. Bias, linearity and stability need their own studies, with reference standards and repeated runs.
    • One study is not lasting proof. NIST treats stability as variation from run to run over time, which only repeated runs can show.
    • Acceptance limits are not universal. Customers and manuals set them, and customer requirements change, as the 2026 gauge R&R clarification noted on IATF's site shows.
    • MSA does not improve the process itself. It shows how much of the observed spread is measurement, before capability figures or control limits are trusted.

    How measurement system analysis relates to other glossary terms

    MSA sits between calibration and the statistics that rely on measured data.

    • Calibration: the comparison with a reference that comes before a gauge study, not instead of it.
    • Statistical process control: the charts whose signals may be measurement noise when the gauge was never studied.
    • Process capability: the indices whose observed spread includes measurement variation.
    • Quality control: the accept and reject decisions that a poor measuring process can get wrong.

    Records and evidence that usually exist for measurement system analysis

    • A study plan naming the gauge, parts, operators, repeats and acceptance criteria.
    • Raw data sheets or files holding every reading.
    • Study reports with variance components, bias and linearity results and the decision.
    • Calibration certificates for the gauges and reference standards used.
    • Control charts of check standards that show stability over time.
    • Customer requirements that set study methods or limits, with the version applied.

    What a software tool needs to capture for measurement system analysis

    • Study designs with parts, operators and repeats, and the measuring order actually used.
    • Raw readings stored with part, operator, trial and time, never only the summary figures.
    • Variance components computed by a named method, with the method and software version kept.
    • Linearity and bias analysis against reference values, with slope, intercept and their uncertainty.
    • Acceptance criteria held per customer or characteristic, not fixed in the code.
    • Links to the gauge's calibration record and to every capability report that used it.
    • Evidence that the calculations were checked against known data before results are relied on.

    Catalog profiles documenting SPC or process-capability records

    Listed profiles meet the rule of the SPC software guide: a matching tag and a capability row for SPC or process capability, which each entry marks as documented, conditional or vendor-stated. That rule does not test gauge studies, so check with your own data that a product stores raw study readings and computes variance components by a stated method.

    Same rule for every product: tagged “Statistical Process Control (SPC)”, “SPC”, “Manufacturing Quality”, “Quality Inspection” or “Quality Management”, with a sourced SPC or process-capability capability in its record, or one the vendor stated to us, labelled as vendor-stated. Documented and conditional capabilities qualify; check the linked scope before treating a module as included.

    Tekmon pays for a sponsored placement (above); its place in this list follows the same rule as every entry. How lists are ordered

    All 4 matching profiles are shown. Documentation review, not hands-on testing or a claim of compliance.

    Related software

    Frequently asked questions

    What is the difference between MSA and gauge R&R?

    Gauge R&R is one study within MSA, covering repeatability and reproducibility. MSA also covers stability over time, bias against reference values and linearity across the range, which NIST's handbook treats as separate analyses.

    What %GRR or ndc value is acceptable?

    This page states no limit. The figures often quoted are usually attributed to AIAG's Measurement Systems Analysis manual, which is paid and was not read, and customers can set their own. Use the criteria your customer or quality plan names, and record them with the study.

    Does calibration replace a gauge study?

    No. In NIST's handbook calibration estimates or removes bias against a reference standard, yet temperature, humidity, operators and wear can still add bias at the workplace. A gauge study is designed to find the bias and variation that appear in use, so the two are done together.

    More quality terms

    Related guides