Guide
ISO 17034 for Reference Material Producers: What CRM Producers, QC Labs and Pharma Standards Teams Actually Have to Evidence
ISO 17034:2016 — General requirements for the competence of reference material producers — is the international standard accreditation bodies use to certify that a producer's reference materials (RMs) and certified reference materials (CRMs) are fit for use in calibration, method validation, quality control and the assignment of values to other materials. It is the standard behind every accredited CRM you buy from NIST, LGC, IRMM, NMIJ, ERM, USP or the national metrology institutes — and behind every in-house CRM a pharma, medical-device, agrochemical or food laboratory produces for internal use. It sits next to ISO/IEC 17025 (testing and calibration laboratories) and ISO/IEC 17043 (proficiency testing providers) in the metrology accreditation stack, and the three are often held together by the same organisation. This guide breaks the standard into the artefacts an accreditation assessor will actually inspect — characterization, homogeneity, stability, metrological traceability, commutability, value assignment, uncertainty budgets and the CRM certificate — the failure modes that appear repeatedly in surveillance visits, and a realistic readiness path. It is written for CRM producers, QC and analytical-chemistry leads, USP/EP/JP reference-standard teams, calibration labs, and any laboratory that produces in-house standards used to make GxP or regulatory decisions.
Start free trial Free trial, no credit card, onboard in days, not months.
What ISO 17034 actually covers (and what it does not)
ISO 17034:2016 sets the management-system and technical-competence requirements an organisation must satisfy to be recognised, by an ILAC-MRA accreditation body, as a competent producer of reference materials. Its scope covers the production of both RMs (sufficiently homogeneous and stable with respect to one or more specified properties) and CRMs (RMs accompanied by a documented certificate, with values established by metrologically valid procedures and traceable to a stated reference, each with an associated uncertainty at a stated level of confidence). The standard does not cover the use of CRMs by the end-user laboratory — that is ISO/IEC 17025 territory — and it does not cover proficiency-testing scheme operation, which is ISO/IEC 17043. The line matters: a QC lab that buys CRMs and uses them is not in scope of 17034; a QC lab that produces an in-house secondary standard used to release product, qualify a method or calibrate an instrument is, and accreditation bodies have made it clear that 'we make it ourselves' does not exempt the work from competence requirements equivalent to 17034. ISO Guides 30 (terms and definitions), 31 (contents of CRM certificates), 33 (use of CRMs) and 35 (characterization, homogeneity and stability) are the technical companions and are referenced normatively or informatively throughout.
Characterization: the technical core of the standard
Section 7 of ISO 17034 covers production planning, characterization, homogeneity, stability, value assignment and the assignment of measurement uncertainty — the technical core of the standard and the bulk of an assessor's time on site. Characterization is the procedure by which the property values of the candidate RM are established, and the standard recognises four broad approaches: (a) a single primary method of demonstrated accuracy applied at one competent laboratory, (b) two or more independent reference methods, (c) one method applied at a network of competent laboratories (inter-laboratory characterization), or (d) a method-specific approach where the property value is intrinsically tied to the method (e.g. a matrix CRM for a regulatory method). Each approach has different uncertainty consequences and different evidence requirements. The recurring failure mode is choosing the inter-laboratory route for cost reasons without adequate participant qualification, then discovering at the value-assignment stage that participant scatter dominates the uncertainty budget and the certified value is no longer fit for purpose. The standard requires the characterization plan to be documented before the work starts, the participants and methods qualified, and the statistical model for combining results pre-specified.
Homogeneity studies and the between-bottle uncertainty
Homogeneity studies establish that the property value does not vary unacceptably between units of the batch (between-bottle) and within a unit at the minimum sample mass the producer specifies (within-bottle). The standard requires a homogeneity study following a documented plan — typically a one-way ANOVA on a representative number of units selected by a defensible sampling design — and the between-bottle component of the uncertainty (u_bb) must be included in the certified uncertainty unless it can be shown to be negligible. ISO Guide 35 is the practical reference for the design, and the analytical method used must have a repeatability low enough to resolve the between-bottle variance — using a method with poor repeatability inflates the apparent u_bb and is one of the most common reasons in-house CRMs fail their first surveillance assessment. Minimum sample mass is the other recurring problem: the certified value applies only at or above the minimum sample mass stated on the certificate, and end-user labs that draw smaller subsamples are no longer working with a traceable CRM. The producer must state the minimum sample mass and the evidence behind it.
Stability studies: short-term, long-term, isochronous design
Stability has two components: long-term stability (the property value over the shelf life under recommended storage conditions) and short-term stability (the property value during transport and at end-user conditions for a limited window). The standard requires both to be evaluated, with the resulting uncertainty components (u_lts and u_sts) included in the certified uncertainty unless shown to be negligible. The recommended design for long-term stability is the isochronous study (Lamberty/Pauwels): units stored under various conditions are returned to a reference condition at staggered intervals and measured in a single analytical run at the end, which eliminates between-run analytical drift from the stability estimate and is widely accepted by accreditation bodies. The classical degradation-rate model (Arrhenius) is acceptable for chemically labile materials with a known degradation pathway but is hard to defend for complex matrices. The recurring failure mode is treating a six-month real-time stability dataset as a five-year shelf-life claim by visual inspection of a trend line — the standard expects a statistical test (typically a t-test on the slope) and an uncertainty component derived from the model, not an inspector's eye.
Metrological traceability and commutability
Metrological traceability — the property by which a measurement result can be related to a stated reference (typically the SI) through an unbroken chain of calibrations, each with stated uncertainty — is required for every certified property value, and the chain must be documented on the certificate. For pure-substance CRMs the chain is usually short and to the SI (mole or kilogram via a primary-method laboratory); for matrix CRMs the chain may run via a higher-order CRM or via a primary reference measurement procedure, and where no SI traceability is possible (e.g. enzymatic activity, immunoassay calibrants) the producer must state the alternative stated reference clearly. Commutability — the equivalence in measurement response between the CRM and routine clinical or environmental samples across a defined set of measurement procedures — is a separate property required for CRMs intended for use as calibrators across multiple methods, and is governed by CLSI EP30-A / IFCC working-group guidance. A CRM that is metrologically traceable but not commutable will produce method-specific bias when used as a calibrator across platforms, and the certificate must be honest about which use cases the material is fit for. Inspectors increasingly ask for a commutability statement on any CRM marketed for inter-method use.
Value assignment and the uncertainty budget
The certified value of a CRM is established by combining the characterization result with the homogeneity and stability uncertainty components, propagating them through a documented model. The standard requires the producer to assign a measurement uncertainty to every certified value, expressed as an expanded uncertainty U at a stated coverage factor k (almost always k=2 for ~95% confidence) and stated in the same unit as the value. The combined standard uncertainty u_CRM is the root-sum-square of the characterization uncertainty u_char, the between-bottle homogeneity uncertainty u_bb and the long-term stability uncertainty u_lts, following the GUM (Guide to the Expression of Uncertainty in Measurement, JCGM 100:2008). The recurring failure mode is omitting one of the three components — typically u_lts because the stability study is still running at certificate issue — and then issuing a certificate with an uncertainty that is structurally too small. ISO 17034 requires the omission to be either justified (component shown to be negligible) or the certificate withheld until the missing component is available; the workaround of issuing a 'provisional' certificate with the missing component to be added later is not accepted by ILAC-MRA assessors.
The CRM certificate (ISO Guide 31) and the product information sheet
The CRM certificate is the legally meaningful artefact the end-user laboratory relies on. ISO Guide 31:2015 sets the minimum contents: identity of the producing organisation, unique identifier, description and intended use of the material, certified property values with their uncertainties and units, the stated reference for traceability, the analytical methods and the statistical model used, the minimum sample mass, the storage conditions, the expiry date or period of validity, the date of certification and the name and title of the signatory. ISO 17034 requires the certificate to be reviewed and authorised by a designated technical authority before issue and a controlled record of every certificate version retained for the lifetime of the material plus a retention period. The recurring failure mode is the certificate that has drifted from the underlying study records — value to two decimal places on the certificate, value to three decimal places in the characterization report, with no documented rationale for the rounding. Assessors trace from the certificate back to the raw data, and any unexplained difference is a finding.
Post-certification surveillance, recalls and the customer-complaint pathway
Production of the CRM is not the end of the obligation. ISO 17034 requires ongoing stability monitoring (post-certification stability) at scheduled intervals across the declared shelf life, with results compared against the certified value and the stability model. If the post-certification data indicate the property value has shifted outside the certified uncertainty, the producer must either re-certify the material at the new value, issue an erratum reducing the shelf life, or recall the affected lots and notify all known users. The standard also requires a customer-complaint pathway and a documented procedure for handling complaints — including investigation, root-cause analysis, corrective action and, where the complaint reveals a wider problem, recall. The recurring failure mode is treating post-certification stability as a paperwork exercise — running the studies, observing drift, and not actioning it because the operational consequences (recall, customer notification, re-certification cost) are large. Assessors look specifically for the link between the stability monitoring output, the management review and the actions taken.
Joint accreditation with ISO/IEC 17025 and ISO/IEC 17043
Most organisations that produce CRMs are also accredited testing or calibration laboratories under ISO/IEC 17025, and many also operate proficiency-testing schemes under ISO/IEC 17043. The three standards are deliberately aligned (impartiality, confidentiality, competence, management-system requirements based on ISO 9001 principles) and joint accreditation is the operational norm — one management system, three technical scopes, one set of internal audits and management reviews, one accreditation body. The economy of effort is real but the trap is using a 17025-style management system without the 17034-specific technical artefacts (production plan, homogeneity study, stability study, traceability chain, CRM certificate, post-certification monitoring). Assessors will explicitly look for evidence that the producer understands the differences — and that the production records would survive an assessor who has read ISO Guide 35 line by line.
In-house CRM production: when accreditation is not the goal
Many QC labs in pharma, medical devices, food and agrochemicals produce in-house secondary standards used for instrument calibration, method qualification or batch release, with no intention to sell them or seek 17034 accreditation. The regulatory expectation is nonetheless that those standards are produced under a system equivalent in rigour to 17034, because the decisions made with them have the same patient-safety, regulatory or commercial consequences as decisions made with an accredited CRM. FDA Form 483 observations, EU GMP Chapter 6 deficiencies and OOS investigations regularly land on in-house reference standards with no homogeneity evidence, no documented stability protocol, no uncertainty budget and a certificate that says only 'characterised internally'. The pragmatic path is to apply the technical sections of 17034 (Section 7) and ISO Guide 35 to in-house production without seeking formal accreditation — production plan, characterization against a higher-order CRM, homogeneity check, stability protocol with scheduled re-checks, uncertainty estimate, and a controlled internal certificate. The result is a standard that survives an FDA pre-approval inspection, an EU GMP audit and a method-transfer dispute with the receiving site, without the cost of accreditation.
Where this lives in V5 Ultimate
The clauses above aren't theoretical — every one maps to a shipped module and an industry profile. Jump to the parts of the product that turn this guide into evidence on a Monday morning.
Modules that do the work
Industries this hits hardest
Frequently asked
Does ISO 17034 replace ISO Guide 34?
Yes. ISO 17034:2016 superseded ISO Guide 34:2009 on its publication and the transition was completed across ILAC-MRA accreditation bodies during 2017–2019. Any accreditation certificate citing Guide 34 today is either historical or in error. The technical companions — ISO Guides 30, 31, 33 and 35 — remain in force.
Do we need 17034 accreditation if we only produce reference materials for internal use?
No — 17034 accreditation is required only for CRMs marketed externally as accredited. But regulators (FDA, EMA, MHRA) expect in-house secondary standards used for GxP decisions to be produced under a system equivalent in technical rigour to 17034. The pragmatic path is to apply Section 7 of 17034 and ISO Guide 35 (production plan, characterization, homogeneity, stability, uncertainty budget, internal certificate) without seeking formal accreditation.
What is the difference between an RM and a CRM?
An RM (reference material) is sufficiently homogeneous and stable with respect to one or more specified properties for its intended use. A CRM (certified reference material) is an RM accompanied by a documented certificate per ISO Guide 31, with property values established by metrologically valid procedures, traceable to a stated reference (usually the SI) and each accompanied by an uncertainty at a stated level of confidence. Every CRM is an RM; not every RM is a CRM.
Does ISO 17034 require commutability?
Not for every CRM — only where the CRM is intended for use as a calibrator across multiple measurement procedures, particularly in clinical chemistry. Commutability is evaluated per CLSI EP30-A / IFCC working-group guidance and is reported on the certificate as a separate statement. A CRM that is traceable but not commutable will still be useful for method validation or quality control within a single method, but should not be used as an inter-method calibrator and the certificate must say so.
How is the uncertainty on a CRM certificate calculated?
The combined standard uncertainty u_CRM is the root-sum-square of the characterization uncertainty u_char, the between-bottle homogeneity uncertainty u_bb and the long-term stability uncertainty u_lts (plus the short-term stability component u_sts where the material is transported under conditions different from recommended storage). The expanded uncertainty U on the certificate is u_CRM multiplied by a coverage factor k (almost always k=2, corresponding to approximately 95% confidence under a normal distribution), and is reported in the same unit as the certified value. The propagation follows the GUM (JCGM 100:2008).
Can we issue a provisional CRM certificate while the stability study is still running?
No — ILAC-MRA accreditation bodies do not accept provisional certificates for accredited CRMs. The certificate must include the long-term stability uncertainty component, or have it shown to be negligible with evidence. The pragmatic options are: issue the certificate with a conservative shelf life supported by the available stability data and revise upward as more data accrue, or hold the lot in inventory until the stability study supports the intended shelf life. The shortcut of 'provisional now, update later' is a recurring assessment finding.
How does ISO 17034 interact with USP, EP and JP reference standards?
USP Reference Standards, Ph. Eur. Chemical Reference Substances (CRS) and JP Reference Standards are produced by pharmacopoeial authorities under their own quality systems, which align with — but are not formally accredited against — ISO 17034. They are accepted as legally binding references for pharmacopoeial assays in their respective jurisdictions. A pharma laboratory using them for compendial methods does not need to re-qualify them; a laboratory using them as the higher-order CRM in the traceability chain of an in-house secondary standard should document the link and treat the pharmacopoeial standard as the stated reference for traceability purposes.
What is the minimum sample mass and why does it matter?
The minimum sample mass is the smallest portion of the CRM at which the certified value and its uncertainty apply, established during the homogeneity study and stated on the certificate. Subsampling below the minimum mass means the within-bottle inhomogeneity is no longer averaged out, the effective uncertainty becomes larger than the certified value indicates, and the result is no longer metrologically traceable in the strict sense. End-user labs that routinely use 5 mg of a CRM with a 50 mg minimum sample mass are operating outside the certified envelope and may produce biased results without realising it.
See it on your shop floor.
Free trial, no credit card, onboard in days, not months.
Related reading
AI-assisted draft·Reviewed against the V5 marketing knowledge base. Spot something off? .