PHARMA LAB · PL-05-019
Calibration, Adjustment and Verification: Differences and Practical Examples

In this article
“The instrument was calibrated” can describe very different activities: comparison with a reference, a coefficient change or a check ending in PASS. To decide whether the evidence serves the laboratory, establish which operation took place and which question it answers.
Calibration, adjustment and verification may belong to the same service, but they are not synonyms. Separating them in the order and report prevents buying a different activity from the one needed or attributing unsupported coverage to a limited test.
1. Calibration: establish the measurement relationship
The VIM describes calibration as a process relating reference values and indications, with associated uncertainties, and using that information to obtain results from indications. Writing “PASS” is therefore insufficient: the data must make the relationship usable under stated conditions.
Results may include errors, corrections, a curve or a table, with uncertainty and an applicable range. Specify the quantity measured, points representing use and configuration tested. Calibrating a disconnected sensor does not automatically describe the complete chain including acquisition and software.
A measured error does not make calibration useless: it is precisely information calibration can provide. Suitability then depends on requirements, uncertainty and how the laboratory uses the results. A certificate does not automatically make every use of the instrument suitable.
2. Adjustment: change the response
Adjustment acts on the system to obtain prescribed indications at reference values. It may address zero, offset, gain or other parameters specified by the method and manufacturer. Record who authorises it, why it occurs, which settings change and at which stage.
The change is not proof of the resulting performance. The VIM states that recalibration is usually necessary after adjustment: define testing relevant to the changed response and intended use. Retain earlier data when available and do not use final results to rewrite the initial state.
Also distinguish adjustment from correcting a result. Applying a calibration-derived correction in controlled processing compensates an estimated effect without necessarily changing the instrument. Loading coefficients that change its indication instead changes the system or configuration. Both require rules, records and verification of implementation, avoiding duplicate corrections.
3. Verification: compare evidence with specified requirements
Verification seeks evidence that an item meets stated requirements. It may use calibration results but also needs a relevant specification, test coverage and an assessment criterion. “Within limit” without the limit and conditions is not a reproducible conclusion.
Where uncertainty matters to the decision, state how it is considered. Agree the rule before observing the outcome rather than selecting it afterwards to obtain a PASS. Distinguish manufacturer specifications from method requirements: their ranges, conditions and tolerances may differ.
This article concerns internal or contractual technical verification. Legal metrology verification follows its own rules, scope and authorised bodies. An internal check is not automatically legal verification; calibration does not replace applicable legal obligations. The CEM distinction between these domains does not transfer Spanish regulations to every laboratory.
4. Original table: input, action, output and limit
| Activity | Input | Action | Output | Does not alone demonstrate |
|---|---|---|---|---|
| Calibration | References, configuration and conditions | Establish the metrological relationship and uncertainty | Results, errors or corrections within tested coverage | Suitability for every use or completed adjustment |
| Adjustment | Observed response, target and authorisation | Change relevant parameters or components | New configuration and intervention record | Final performance across the entire range |
| Verification | Requirement, evidence and decision rule | Assess fulfilment of the requirement | Bounded, justified conclusion | Characterisation of untested points or functions |
| Result correction | Model, valid correction and original data | Compensate an estimated systematic effect | Corrected result with relevant uncertainty | Physical instrument modification or absence of uncertainty |
Maintenance, repair and qualification may interact with these activities but answer other questions. Cleaning is not automatically adjustment; a new component is not automatically qualified; qualification does not remove the need for relevant metrological evidence.
5. Read the command, understand the operation
A “calibration”, “auto-cal” or “standardisation” button does not identify the whole metrological activity. Consult the manual: does it acquire readings without changing them, calculate a relationship, overwrite coefficients, check a specification or combine operations? Describe the actual sequence.
In Italian, the metrological term corresponding to calibration is “taratura”; retain the original command name when needed for identification. Do not tidy terminology in the record by erasing what the software actually did. Link the command, version, configuration, inputs, changes and retained outputs.
A successful self-diagnostic may concern internal electronics or functionality without comparing the quantity with an external reference. Even where an internal reference exists, assess stability, traceability and coverage: the function’s name alone does not demonstrate the whole chain. Ask the provider for relevant output examples, not just the service package’s commercial name.
6. Simulated case and service order
A laboratory uses a thermometric chain between 20 and 80 °C. The technician adjusts the offset at 0 °C and subsequently checks the same point, with a favourable outcome under the documented criterion. The record demonstrates the intervention and the check at 0 °C, not performance throughout the use range.
The responsible person requests results at points representing use, uncertainty, the complete configuration and test conditions, followed by assessment against laboratory requirements. The plan needs technical justification; it does not follow automatically from the number of points in this example. Zero adjustment cannot be assumed to correct sensitivity, nonlinearity or other contributions.
A more precise order distinguishes calibration in the agreed condition; adjustment if authorised; calibration after intervention; and verification against identified requirements and a decision rule. Specify deliverable data, who assesses doubtful outcomes and who authorises use. Agree normal instrument preparation conditions without automatically treating them as repairs.
The laboratory reviews what actually occurred, keeps results and interventions separate and decides on use. For chronology, read As Found and As Left; for uncertainty, consult the measurement uncertainty budget. Return to the Calibration and Metrology hub.
7. Sources and limits
Checked on 1 October 2026. The table, example order and case are original; the temperatures are not mandatory points prescribed by a standard.
- VIM, third edition, calibration §2.39, adjustment §3.11, verification §2.44 and correction §2.53: complete official entries consulted; applied paraphrases, not reproduced normative text.
- CEM, calibration and verification differences: complete institutional page, no visible revision date; used for conceptual distinctions, not advice on the reader’s legal obligations.
- Accredia, certificate Q&A, 27 September 2023: PDF consulted, answers on repair, normal preparation, data and Italian terminology; institutional explanations, not the full ISO/IEC 17025 text.
Continue exploring
PL-05-020
Intermediate Checks: Verifying Instruments Between Calibrations
A useful check detects a change relevant to use. Define what it monitors, which errors it may share with the instrument and what remains outside its coverage.
Read the articlePL-05-018
As Found and As Left: Meaning and Use of Calibration Results
A documented sequence shows which data describe the instrument before and after service, without asking final results to reconstruct a history they cannot establish.
Read the articlePL-05-016
Calibration Intervals: Risk, Performance History and Review
Three passing certificates do not automatically demonstrate long-term stability. Separate comparable data, interventions and checks before changing a due date.
Read the article


