Setting and justifying calibration intervals

The widespread assumption that an instrument must be calibrated 'once a year' has no normative basis. Clause 6.4 of ISO/IEC 17025 requires that equipment is calibrated where measurement accuracy or uncertainty affects the reported result, and that a calibration programme exists. What interval that programme sets is left open.

That is not an omission but a consequence: a torque wrench used daily at the limit of its range drifts differently from an identical one taken out of a cabinet twice a year for a check measurement. A blanket interval would be too long for one and a waste of money for the other.

The consequence for the laboratory: the decision is yours, and so is the burden of proof. An audit does not ask whether you have an interval, but why it is this one and how you would notice it was too long.

Six factors determine the interval. They can be worked through for any instrument in a few minutes and should appear in the justification:

• Manufacturer's recommendation — the starting point, never the complete justification. The manufacturer does not know your operating conditions.
• Frequency and intensity of use — daily use at the limit of specification justifies shorter intervals than occasional use in mid-range.
• Operating and environmental conditions — vibration, temperature cycling, humidity, contamination and transport between sites all accelerate drift and misadjustment.
• Required measurement uncertainty relative to the instrument's specification — the narrower the margin between permissible deviation and actual accuracy, the less drift the interval can absorb.
• Stability from your own history — previous calibration results for the same instrument or the same model are the most defensible source once you have them.
• Consequences of a wrong measurement — if a detected deviation would force re-evaluation of everything produced since the last valid calibration, a shorter interval is the cheaper option.

The internationally used reference for determining and adjusting calibration intervals is the joint document ILAC-G24 / OIML D 10. It describes five methods, which can also be combined.

1. 1

   Staircase (automatic) adjustment

   If the instrument is found within tolerance at calibration, the interval is extended by a fixed step; if it is out of tolerance, the interval is shortened. Simple to operate and well suited to large populations of similar instruments. It requires that you define the step size and the upper and lower bounds in advance rather than deciding case by case.

2. 2

   Control chart

   Deviations for selected characteristics are plotted over time across calibrations. Scatter and drift allow you to project when the tolerance limit would be reached, and the interval is chosen so that point is safely not crossed. More effort, but methodologically the cleanest option and the easiest to defend in an audit.

3. 3

   In-use time instead of calendar time

   The interval is expressed in operating hours, measurement cycles or number of tests rather than months. Suited to instruments with strongly varying utilisation. It requires that usage is actually counted — the most common reason this method fails in practice.

4. 4

   In-service checking (black-box testing)

   Between calibrations the instrument is checked against a known reference or check standard. This extends the calibration interval itself without increasing risk, because misadjustment is spotted early. Particularly worthwhile for instruments where calibration is expensive or causes long downtime.

5. 5

   Statistical approaches

   Across an entire family of instruments, the proportion found out of tolerance at calibration is evaluated, and the interval is set so that proportion stays below a defined threshold. Needs a sufficiently large population and only pays off from a certain fleet size.

Every adjustment method needs data that a newly purchased instrument does not yet have. A conservative sequence works well for the start:

1. 1

   Begin with the manufacturer's recommendation

   It is the only available reference point and is accepted as a starting point — as long as it is documented that it was the starting point and not the end of the reasoning.

2. 2

   Shorten where operating conditions are harsher

   Rough use, tight tolerances or high consequential cost of a wrong measurement justify halving the first interval. Extending happens only once you have your own data, never at the outset.

3. 3

   Review after the second and third calibration

   From the second data point onward you can see the actual drift. This is where the interval is either confirmed or adjusted for the first time — and exactly the review that is missing from most calibration plans.

4. 4

   Use similar instruments as a reference

   If you already run identical models under comparable conditions, their history is a better basis than the manufacturer's data sheet. The same applies to a newly purchased replacement.

The audit question is rarely 'how long is your interval?'. It is whether the interval was a traceable decision and whether it is alive. Concretely, you should be able to produce:

1. 1

   The interval is never touched again

   The value set once still sits unchanged in the list five years later, although usage and location have changed. The standard does not require adjustment for its own sake — but it does require review, and the absence of one is demonstrable.

2. 2

   The due date is tracked as a date, not as a state

   A list of due dates answers 'which instruments are due today?'. It does not answer 'was this instrument validly calibrated on 14 March, when we produced report 2026-0412 with it?' — and that is the question actually asked in an audit.

3. 3

   Extension without a data basis

   An interval is extended because calibration is expensive or the instrument is needed, not because the calibration results support it. This is the variant an audit spots fastest, because the justification is missing.

Calibration intervals are not a documentation problem but a linking problem. The interval itself goes into a spreadsheet quickly. What gets hard is the connection between instrument, calibration status at a particular point in time, and the tests performed with it.

• Every instrument carries its calibration history with results, not only the next due date.
• Calibration status can be queried for a point in time, so it can be established retrospectively which tests are affected.
• Due dates become visible before they lapse, not after — with enough lead time to commission the work.
• Blocked instruments can no longer be selected in running tests, instead of merely being marked as blocked on a list.
• The justification for the interval lives with the instrument, not in a separate document that disappears with the next staff change.

TestLabIQ models these links: equipment with calibration history and due dates, tests recording the instrument actually used, and an audit log over every change. Whether that fits your equipment fleet is fastest to establish in a demo using your own examples.