Calibration intervals can be lengthened, not only shortened, when as-found data and built-in instrument diagnostics show the device is staying inside its tolerance band, with ILAC-G24:2022/OIML D 10:2022 and ISO/IEC 17025:2017 setting the documented risk-based framework [S3][S4].
Across a 10-year internal dataset of 381,916 calibrations, intervals that were set once and never revisited are the leading cause of both surprise out-of-tolerance findings and wasted calibration spend, an audit pattern that is now showing up in pharmaceutical, biomedical, and process-plant assessments under [S5] (2026-07). The key shift in 2026 is that the equipment owner, not the calibration lab, owns the interval decision; the lab supplies as-found/as-left evidence and uncertainty, and the quality system turns that evidence into a defensible schedule [S3][S5].
What "extending the interval" actually means in 2026 practice
An interval extension is a written, data-supported decision to lengthen the time or usage count between two successive traceable calibrations of an instrument whose output feeds a reportable result, governed by ISO/IEC 17025:2017 Section 6.4 and supported by ILAC-G24:2022/OIML D 10:2022 [S3][S4]. The interval is expressed either as elapsed time (e.g. "every 24 months") or as a usage trigger (e.g. "every 5,000 cycles" or "after 500 hours of run time"), whichever better tracks the device's actual drift behavior [S3]. It is not the same as a maintenance schedule (cleaning, filter changes) or a verification check (a quick before-use spot-check); a calibration interval specifically governs when a full, traceable calibration against certified reference standards is required, typically by an ISO/IEC 17025-accredited provider or a qualified in-house metrology function [S3]. NIST GMP 11 (2019) is the corresponding U.S. baseline practice, and it caps legal-metrology intervals at 10 years unless a detailed technical and statistical assessment of historical data, control charts, check standards, and proficiency tests is approved by the NIST Office of Weights and Measures [S1].
When an extension is defensible, and when it is not
Extensions are defensible when the instrument has a documented history of multiple consecutive in-tolerance as-found readings, the operating environment is no harsher than the manufacturer's assumption, and the resulting measurement risk is acceptable to the user [S3][S5][S7]. Tektronix guidance, dated 2025-10, states explicitly that an instrument which has consistently remained in tolerance over multiple calibration cycles may be considered for an extended interval, provided that stability, usage, and environmental conditions are reviewed [S7]. A practical numeric gate, used by the Techmaster reliability-based program, is an End-of-Period Reliability (EPR) target of 85 to 95 percent; intervals are lengthened when measured reliability stays above the chosen target and shortened when out-of-tolerance events cluster [S5]. Extensions are not defensible when legal or industry rules fix the interval, when the instrument has had any in-tolerance excursion that was traced to a real drift mechanism rather than a handling shock, when intermediate check standards show upward drift, or when the device has been repaired, had a major firmware change, or been moved to a harsher environment [S3][S4][S8].
The role of instrument diagnostics and self-diagnostics
Modern electronic test and process instruments increasingly embed built-in circuitry and firmware that monitor internal state and self-adjust to stay inside a self-adjustment range, and that built-in diagnostic data is what makes 24- and 36-month intervals technically defensible rather than merely administrative [S2]. Keysight's published 2010-2012 dataset shows the share of new products carrying 24- and 36-month recommended intervals rising versus the traditional 12-month default as self-monitoring circuitry matured, and that trend has continued into 2026 with process transmitters and lab instruments alike [S2]. For pressure and flow loops, the same logic shows up as on-board diagnostic coverage in pressure transmitter and flow meter firmware, where HART or FOUNDATION Fieldbus status bytes report loop integrity, sensor health, and calibration date, giving the reliability engineer an independent evidence stream to add to the as-found calibration record. The diagnostic stream does not replace a traceable calibration; it supplements it, and any extension that relies on diagnostics must still be backed by periodic in-tolerance as-found results from a competent provider [S3][S5].
Decision criteria: time-based, usage-based, and reliability-based methods side by side
ILAC-G24:2022/OIML D 10:2022 lists five accepted methods for setting and adjusting intervals, and the right choice depends on the asset class, data availability, and consequence of a missed reading [S4][S5]. The table below lines the three most commonly used in 2026 up against four decision criteria that engineers and auditors actually check.
Straight comparison: time-based default vs usage-trigger vs reliability-based adjustment, on data needed, best fit, typical use case, and audit posture [S4][S5]. Time-based default: needs only manufacturer recommendation, fits low-criticality general lab equipment, used as administrative starting point, audit posture is acceptable but weak. Usage-trigger: needs run-hour or cycle counter, fits high-utilisation production assets like autoclaves and process sensors, used where drift is dominated by wear, audit posture is good when the counter is verified. Reliability-based adjustment: needs as-found history from 3+ cycles plus EPR, fits stable process instrumentation such as analytical instrument channels and reference standards, used when extending beyond manufacturer default, audit posture is the strongest because the data is on the certificate [S4][S5]. The reliability-based method is the one the 2026 guidance consistently points to when an extension beyond the manufacturer's number is on the table, with the EPR target of 85 to 95 percent as the gate and NCSLI RP-1 as the U.S. implementation reference [S5].
Documenting an extension: what an auditor actually wants to see
An extension decision must be supported by a written rationale, the as-found data series, the EPR computation, the manufacturer's recommendation, the relevant standard clause, and a review date or trigger for re-evaluation [S3][S4][S8]. The 2026 SciSafety Alliance and Micro Precision guidance both stress that the standard does not hand the lab a table of fixed intervals; it requires the lab to justify whatever interval it chooses and to revisit that justification as calibration history accumulates, with industry-specific obligations (FDA, GMP, AS9100D, IATF 16949) checked before any change to an established schedule [S4][S8]. A 2023 IOP Conference Series paper (Miqueleti, cited by 2) goes further and presents a mathematical model that assigns a calibration value and uncertainty for extrapolating the period defined in the calibration plan, which is one of the few peer-reviewed approaches to interval extension grounded in measurement-uncertainty budgets rather than rule-of-thumb factors [S9]. For temperature loops, the same logic shows up in temperature calibration bath work, where bath stability, uniformity, and loading tolerance become the as-found data feeding the EPR calculation.
Failure modes and limits: when to shorten, not extend
Extensions fail in three predictable ways: an out-of-tolerance event that is detected only after a long interval, a measurement-uncertainty growth that pushes the result outside the user requirement, and a regulatory change that re-fixes the interval from above [S3][S8]. The corrective action in each case is to shorten the interval, not to retract the extension, and to add an intermediate verification check on the affected range until two or three in-tolerance cycles have been recorded again [S5][S7]. An extension should also be re-examined after any of: a documented repair, a firmware or software change that affects the measurement path, a relocation to a harsher environment, or a change in the measurement uncertainty the user requires [S3][S4]. Equipment owners who want a stable reference frame can use the process calibration workflow as a checklist: confirm range, confirm uncertainty budget, confirm environmental class, then make the call on the interval, in that order.
Sourcing, traceability, and the standards stack that governs the decision
The standards stack for 2026 is: ISO/IEC 17025:2017 (Section 6.4) for the competence and equipment requirements, ILAC-G24:2022/OIML D 10:2022 for the recalibration-interval methods, ANSI/NCSL Z540-1 for the U.S. calibration-system framework, NIST GMP 11 (2019) for U.S. laboratory practice, and NCSLI RP-1 as the implementation reference for reliability-based interval adjustment [S3][S4][S5]. Manufacturer recommendations remain the starting point, with 12 months still the most common default but with 24- and 36-month intervals increasingly specified on instruments with built-in self-monitoring, and 12, 24, and 36 months all appearing in published OEM interval tables [S2]. The data signal to track going into late 2026 is the share of new process-instrument SKUs that publish a 36-month recommended interval in their datasheets; the procedural signal is whether accreditation bodies continue to accept ILAC-G24:2022 reliability-based extension as the primary evidence basis or shift toward fixed regulatory tables for specific asset classes.
See also our earlier report, Expansion Joint: Why a Waterstop Plus a Surface Sealant Is the Default.