How Should You Set a Calibration Interval?

Fixed annual calibration is common because it is simple to administer, aligns with many supplier recommendations and provides a predictable workload. It is not, however, automatically the right interval for every instrument.

A calibration interval should provide reasonable confidence that equipment remains fit for its intended measurement task between calibrations. The appropriate period depends on how the instrument performs in service, how it is used, and what happens if it drifts beyond an acceptable limit without being detected.

The requirement is a programme, not a universal period

ISO/IEC 17025:2017, clause 6.4.7, requires the laboratory to establish a calibration programme that is reviewed and adjusted as necessary to maintain confidence in the status of equipment. It does not prescribe a 12-month interval.

This distinction matters. An annual label may be a valid output of a well-designed programme, but it is not evidence that the selected period is technically suitable. A defensible programme identifies which equipment needs calibration, intermediate checks or other controls, and defines how evidence will trigger interval review.

The interval also concerns more than the instrument itself. A pressure controller used as a working standard, a digital multimeter supporting product release, and a handheld thermometer used for indicative warehouse checks can have very different control needs even where their nominal specifications appear similar.

Start with the intended use and permissible drift

The key question is not, “How often do we normally calibrate this model?” It is, “How much unrecognised change can this measurement process tolerate?”

Start by defining the instrument's intended use, measurement range and required capability. Then identify the maximum error or change in indication that can be accepted in service. This may be derived from:

• a product or process tolerance;
• a test method requirement;
• the required test uncertainty ratio or other capability target;
• the instrument manufacturer's specification; or
• an internal acceptance limit established through technical evaluation.

The acceptance limit for calibration should not be confused with the product specification limit. For example, a torque wrench used to apply a critical assembly torque may need to remain well within its stated maximum permissible error so that process variation and measurement uncertainty do not consume the full product tolerance.

Where a conformity decision is made on calibration results, the laboratory or equipment owner should also define the decision rule. The reported measurement result, its uncertainty, the acceptance limits and any guard band are separate elements. A tighter acceptance limit can reduce the chance of accepting an out-of-tolerance instrument, but may also lead to more rejected instruments. That is a risk decision, not simply an administrative setting.

Use history to establish the first interval and improve it

For new equipment, there may be little internal evidence. A manufacturer recommendation, the interval used by an experienced calibration provider, comparable equipment history and the criticality of the intended use are reasonable starting inputs. In higher-risk applications, an initially conservative interval is usually preferable until performance data are available.

Once an instrument has been calibrated more than once, its history should become the main source of evidence. Review more than pass or fail status. Useful information includes:

• as-found error at each test point;
• change in error since the previous calibration;
• direction and consistency of drift;
• adjustment or repair history;
• whether the instrument was used near the limits of its range;
• environmental exposure, transport and handling; and
• results of intermediate checks.

An instrument that repeatedly returns with small, stable as-found deviations may justify a longer interval, provided the projected drift remains acceptable for its intended use. Conversely, an instrument that passes calibration but shows steadily increasing error may warrant a shorter interval before it actually fails.

For example, a reference balance may show a repeatable zero and sensitivity shift equivalent to 0.4 mg over 12 months. If the maximum permitted in-service change is 2 mg and routine check weights provide effective surveillance, extending to 18 months may be technically supportable. If the same balance supports release testing with little tolerance available, the organisation may retain 12 months or shorten the interval despite the favourable history.

Consider usage and exposure, not time alone

Calendar time is an imperfect proxy for deterioration. Some equipment drifts principally with operating hours, switching cycles, mechanical loading or exposure to adverse conditions.

A portable pressure gauge used daily in field service, subjected to vibration and occasional overpressure, should not generally be managed in the same way as an identical gauge used monthly in a controlled laboratory. Similarly, a micrometer used on a production line may require more frequent checks for wear, damage and zero shift than one retained as a lightly used inspection reference.

A practical programme can use different triggers alongside a calendar due date, such as:

• a maximum number of operating hours or cycles;
• calibration after repair, overload, damage or suspected misuse;
• calibration before a critical campaign;
• shortened intervals after an adverse trend; and
• reinstatement only after satisfactory evaluation following an environmental excursion.

These triggers must be visible in the equipment management process. An interval policy is ineffective if a damaged instrument can remain in use until its next scheduled date.

Intermediate checks can support, but not automatically replace, calibration

Well-designed intermediate checks provide evidence that an instrument remains under control between calibrations. They are particularly valuable for reference standards and frequently used equipment.

The check must be sensitive to the failure modes that matter. A daily check of a pipette at one nominal volume may detect gross change, but it may not reveal performance degradation across the full working range. A resistance check on a temperature indicator may confirm part of the measuring chain while leaving the probe response and immersion effects unexamined.

Define the check standard, method, acceptance criteria, frequency, action limits and response to failure. Ensure the check standard has suitable metrological traceability and capability for the purpose. Trend the results where practical. A sequence of checks within limits can still indicate a developing shift that should prompt review.

Review intervals after nonconforming results

An out-of-tolerance as-found result is not only an equipment event. It may affect the validity of measurements made since the last known acceptable condition. The extent of retrospective evaluation depends on the magnitude and likely onset of the error, the affected measurement range, available intermediate-check data and the significance of the decisions supported by the instrument.

The investigation should also ask whether the interval remains suitable. Shortening it may be appropriate, but the cause matters. A failure caused by accidental damage may require better handling controls rather than a permanently shorter interval. Recurrent drift may indicate an unsuitable instrument, inadequate environmental control, excessive use or an unrealistic acceptance limit.

An annual interval is often a reasonable starting point. It becomes a defensible interval only when the organisation can explain why that period, together with its intermediate controls and response to change, provides adequate confidence for the actual measurement risk.

References

• ISO/IEC 17025:2017, *General requirements for the competence of testing and calibration laboratories*, clause 6.4.7.
• ILAC G24:2022 / OIML D 10:2022, *Guidelines for the determination of calibration intervals of measuring instruments*.
• JCGM 106:2012, *Evaluation of measurement data - The role of measurement uncertainty in conformity assessment*.