ISO 17025 Method Selection, Verification & Validation
Overview
Method validation is one of the most technically demanding requirements in ISO/IEC 17025:2017. Clause 7.2 requires laboratories to demonstrate that every method they use is fit for its intended purpose – but the depth of evidence depends on the method’s origin. A standard EPA method used as published requires verification (confirming your lab can meet the stated performance). A standard method you have modified requires partial validation of the affected parameters. A lab-developed method requires full validation with a comprehensive set of performance characteristics.
This skill handles all three tracks with precision. It builds a master validation procedure that defines how your lab classifies methods, selects performance characteristics, designs experiments, sets acceptance criteria, and documents fitness-for-purpose conclusions. For each specific method, it produces protocol and report templates with the correct characteristic set for the track. The extension tools enforce minimum characteristic counts: at least 3 for verification, 2 for partial validation of affected parameters, and 5 for full validation.
Critically, the skill distinguishes measurement uncertainty (Clause 7.6) from validation parameters. Precision and bias data generated during validation are inputs to the uncertainty budget, not validation characteristics themselves. The iso17025-measurement-uncertainty skill handles the Clause 7.6 estimation.
What It Covers
- Method selection hierarchy – documenting the Clause 7.2.1.1 rationale (international standard, reputable body, peer-reviewed, or lab-developed)
- Verification of standard methods – confirming the lab can meet published performance (trueness/bias, precision, LOD/LOQ if claimed); minimum 3 characteristics
- Partial validation (Cl.7.2.1.4) – re-verifying only the parameters affected by a documented deviation from a standard method
- Full validation – comprehensive fitness-for-purpose demonstration for non-standard or lab-developed methods; minimum 5 characteristics (selectivity, LOD, LOQ, linearity, accuracy/trueness, precision, robustness)
- CRM/reference material strategy – using certified reference materials for accuracy assessment per ISO Guide 33 / ILAC P10, or justified alternatives
- Acceptance criteria – pre-defined criteria traceable to fitness-for-purpose requirements before running experiments
- Experimental design – matrices, concentration levels, replicates, days for intermediate precision
- Statistical methods – ANOVA for intermediate precision, Mandel test for linearity, t-test for bias, Grubbs/Cochran for outliers
- Fitness-for-purpose conclusion – per-characteristic PASS/FAIL and overall suitability statement for the defined scope
- Uncertainty input linkage – documenting how precision SD and bias feed the Clause 7.6 measurement uncertainty budget
Example Conversation
You: We need to verify EPA 200.8 for metals by ICP-MS at LabTech Analytical. We’re running it exactly as published – drinking water and groundwater, 23 elements, range 0.5 to 5000 ug/L. This is for our A2LA accreditation.
Agent: Since you’re using EPA 200.8 as published without modifications, this is a standard method verification (Clause 7.2.2). I need to confirm three things:
- Does EPA 200.8 state performance claims for bias, precision, and detection limits? (Most EPA methods do in the QC sections.)
- What ICP-MS instrument are you using?
- Do you have access to CRMs for metals in water – for example, NIST SRM 1643f (Trace Elements in Water)?
You: Yes, Table 6 in EPA 200.8 has MDLs and the QC sections specify spike recovery ranges. We’re running on an Agilent 7850 ICP-MS, installed last year. We have NIST 1643f and ERA WatR metals PT standards.
Agent: Perfect. For verification, the minimum is 3 characteristics. I’ll design the protocol around:
- Accuracy/trueness – analyze NIST 1643f at least 7 times; compare mean vs. certified values using a t-test
- Precision (repeatability) – calculate RSD from the 7 CRM replicates; must be within EPA 200.8 Table 4 limits
- LOD/LOQ – MDL study per 40 CFR Part 136 Appendix B (7 replicates at 1-5x the expected detection limit)
The acceptance criteria will be: CRM recovery within the certified uncertainty range, repeatability RSD below the method-stated limits, and MDLs at or below the EPA 200.8 published values.
You: That works. But we also want to add soil matrices later – EPA 200.8 doesn’t specifically cover soils, we’d be using the acid digest from EPA 3050B first. Does that change things?
Agent: Yes – once you add a matrix (soil digests) not in the original EPA 200.8 scope, that becomes an extension requiring at least a partial validation. You would need to demonstrate:
- Accuracy in the soil matrix (spike recoveries into representative soil digests, or analysis of a soil CRM like NIST SRM 2709b San Joaquin Soil)
- Precision in the new matrix
- Matrix effects assessment (comparing slopes of calibration and standard addition curves)
For now, I’ll build the drinking water/groundwater verification protocol. We can add the soil partial validation as a separate protocol later.
I’ve drafted the verification protocol. Running
method_validation_protocol_checker…Result: PASS
- Track detected: Standard method – Verification (Cl.7.2.2)
- Method type: Quantitative
- Characteristics found (3/7): accuracy/trueness, precision, limit of detection (LOD)
1 warning: Statistical methods mentioned but not explicitly linked to characteristics. I’ll add: “t-test is used for bias assessment against NIST 1643f certified values; RSD calculation for repeatability per EPA 200.8 Table 4.”
After correction – PASS with no warnings. Protocol saved.
Sample Output Excerpt
The skill produces a master validation procedure plus protocol and report templates. Here is a representative excerpt from a verification protocol for EPA 200.8.
Method Verification Protocol: EPA 200.8 Rev 5.4 – Metals by ICP-MS
1. Method Identification
| Field | Value |
|---|---|
| Method reference | EPA 200.8 Rev 5.4 (Determination of Trace Elements in Waters and Wastes by ICP-MS) |
| Validation track | Standard method verification (Cl.7.2.2) |
| Method type | Quantitative |
| Selection rationale (Cl.7.2.1.1) | Tier 1 – EPA-promulgated method for compliance monitoring under SDWA and CWA |
2. Scope of Verification
| Parameter | Value |
|---|---|
| Measurand(s) | 23 elements: As, Ba, Be, Cd, Cr, Co, Cu, Pb, Mn, Mo, Ni, Se, Ag, Tl, V, Zn, Al, Sb, B, Fe, Li, Si, Sn |
| Matrix | Drinking water, Groundwater |
| Concentration range | 0.5–5000 ug/L |
| Instrument | Agilent 7850 ICP-MS (S/N: MY22140021) |
3. Performance Characteristics
| # | Characteristic | Acceptance Criterion | Experimental Design |
|---|---|---|---|
| 1 | Accuracy/trueness | CRM recovery within certified uncertainty range | 7 independent preparations of NIST SRM 1643f; t-test against certified values |
| 2 | Precision (repeatability) | RSD below EPA 200.8 Table 4 limits per element | Calculated from the 7 CRM replicates |
| 3 | LOD (MDL) | MDL at or below EPA 200.8 published values | 7 replicates at 1–5x expected detection limit per 40 CFR Part 136 Appendix B |
4. Reference Material Strategy
- Primary CRM: NIST SRM 1643f (Trace Elements in Water) – certified values for 26 elements with stated expanded uncertainty (k=2, 95% confidence)
- Secondary check: ERA WatR PT standard analyzed alongside CRM preparations to confirm performance extends to an independent source
- Traceability: NIST SRM 1643f certificates trace to SI units via NIST primary reference standards
Extension Tools
method_validation_protocol_checker
Validates a method validation or verification protocol for Clause 7.2 compliance. Applies track-aware rules based on whether the method is standard (verification), modified (partial validation), or non-standard/lab-developed (full validation).
| Check | What It Validates | Severity |
|---|---|---|
| Method identification | Standard reference or lab designation present | ERROR if missing |
| Selection hierarchy (Cl.7.2.1.1) | Method selection rationale documented | WARNING if missing |
| Validation track | Track identified (verification / partial / full validation) | WARNING if ambiguous |
| Deviation documentation (Cl.7.2.1.4) | For partial validation: deviation described, justified, affected parameters identified | ERROR if missing for partial track |
| Validation objective | Purpose and scope statement | ERROR if missing |
| Scope definition | Matrix, measurand, and concentration range specified | ERROR per missing element |
| Performance characteristics | Minimum count met (3 verification, 2 partial, 5 full) | ERROR if insufficient |
| CRM/reference material | Reference material strategy for accuracy | WARNING if accuracy claimed but no CRM approach |
| Acceptance criteria | Pre-defined criteria for each characteristic | ERROR if missing |
| Experimental design | Replicates, days, concentration levels, matrices | ERROR if missing |
| Statistical methods | Statistical approach defined and linked to specific characteristics | ERROR if absent; WARNING if not linked |
| Range warning | Flags “range” if listed as independent characteristic (it is derived from LOD/LOQ + linearity) | WARNING |
validation_parameter_validator
Validates a completed validation report for Clause 7.2 compliance. Confirms all planned characteristics were evaluated and documented.
| Check | What It Validates | Severity |
|---|---|---|
| Characteristics evaluated | Minimum count met per track | ERROR if insufficient |
| Results vs. criteria | Explicit comparison of results against acceptance criteria | ERROR if missing |
| Per-characteristic conclusion | PASS/FAIL stated for each characteristic | ERROR if missing |
| CRM results | CRM or spiked-matrix recovery documented for accuracy | ERROR if accuracy evaluated without reference material |
| Fitness-for-purpose | Overall suitability conclusion for defined scope | ERROR if missing |
| Scope in conclusion | Matrix, measurand, range specified in conclusion | WARNING if generic |
| Authorization | Approval with name, role, and date | ERROR if missing; WARNING if incomplete |
| Uncertainty inputs (Cl.7.6) | Precision/bias data identified as uncertainty budget inputs | WARNING if not documented |
Getting Started
Start by activating the ISO 17025 Method Selection, Verification & Validation skill. The agent will ask which method you want to validate and determine the correct track (verification, partial validation, or full validation).
Have this information ready:
- The method reference (e.g., EPA 200.8, ASTM D5907, SM 4500-H+ B) or a description of your lab-developed method
- Whether you are using the method as published, with modifications, or it is laboratory-developed
- The matrix types you are testing (drinking water, groundwater, soil, etc.)
- The measurands/analytes and their concentration ranges
- Available certified reference materials (CRMs) and PT standards
- Any regulatory requirements that drive acceptance criteria (EPA, state drinking water program, client specifications)
- Your instrument make and model
If you have already completed the ISO 17025 Laboratory Profile skill, the agent will pull method references and scope information automatically. Measurement uncertainty estimation is handled separately by the ISO 17025 Measurement Uncertainty skill – validation data (precision SD, bias) feed directly into that budget.