What does USP <1058> require for analytical instrument qualification, and what does the proposed revision change?

1What USP <1058> requires, and what it does not.

USP <1058> is a general chapter numbered above 1000, a range USP intends “to provide general guidance for stakeholders”. Its revision briefing puts it plainly: “The decision of ‘how it needs to be done’ is the prerogative of individual organizations.”¹ The official chapter, last revised in 2017, recommends “an integrated approach, based upon a risk assessment” to show fitness for purpose, and covers “any apparatus, equipment, instrument, or instrument system used in pharmacopeial analyses”.² Agilent told USP’s June 2026 workshop that the chapter “intentionally avoids prescribing exact tests/acceptance limits”.³

The obligations come from GMP. <1058> is the method many laboratories use to meet them; USP calls it “widely regarded and used globally by regulators, suppliers, and industry”.¹

The tests and limits come from the general chapter for each technique, numbered below 1000, and from the laboratory’s own methods. At the workshop, Jean-Marc Roussel of USP’s <1058> subcommittee took UV-visible limits from <857>: wavelength accuracy within ±1 nm in the 200 to 400 nm region, absorbance accuracy ±0.010 and precision 0.0050. A reference material certified at 0.4702 read a mean of 0.4660, a difference of 0.0042, with a standard deviation of 0.0024: both within limits. His second example took balance and flask acceptance from <41> and <31>,³ and system suitability tests come from chapters such as <621>.¹² In Europe, the official medicines control laboratories’ equipment qualification guideline, PA/PH/OMCL (08) 73 R7, which other laboratories may use voluntarily, has annexes from chromatographs and spectrophotometers to balances, pH meters and pipettes; its liquid chromatography annex gives typical limits such as flow rate accuracy ±5.0% and column oven accuracy ±2 °C.¹³¹⁴

2Groups A, B and C.

The official chapter, open in full only to USP–NF subscribers, sorts equipment into Groups A, B and C and, in Agilent’s summary, “expects effort to increase with instrument/system complexity and measurement criticality”.³ Eli Lilly and Bristol Myers Squibb showed the workshop how they apply them:¹²¹⁵

The revision published for comment in Pharmacopeial Forum 51(2), March 2025, names eight subgroups, A1 to C3. In USP’s workshop slides, classification “is based upon User intent and dependent on actual way of working”: define the intended use, set aside anything not GMP relevant for the inventory only, then classify.³

The same unit can land in different groups. USP’s example is an ultrasonic bath. Used to dissolve samples with completion judged by eye, it is A1. If the method needs 40 ± 2 °C and 50 watts at 45 kHz for 15 minutes, time, temperature, ultrasonic power and frequency need calibrating, and it is B1. Built into an automated sample preparation system, with time and temperature monitored through application software, it is C1 or possibly C2.³ Companies also draw lines differently: Lilly puts a keypad-operated HPLC connected to a data system in Group C, while BMS keeps an HPLC with no separate software in its B2.¹²¹⁵ Either can be defended if the rationale is written down and applied consistently.

3Instrument classification and qualification record.

One record per instrument answers most inspection questions. Each field meets a requirement:

3.1Preparing for a pre-approval inspection.

FDA’s Compliance Program 7346.832 has investigators compare raw data such as chromatograms with the application’s summary data, naming stability, dissolution, content uniformity and API impurity data as good candidates.¹⁸ Be ready with:

1. The instrument list, with each instrument’s classification and the application data it produced.
2. Qualification and calibration status on the dates those data were made, with any failed checks and their impact assessments.
3. The raw data, audit trails and system suitability results behind each summary table, retrievable on request.

4Example: classifying and qualifying a laboratory’s instruments.

This example is invented: the laboratory, instruments and results are illustrative, and its classification calls are judgements, not USP’s. Cresswold Analytical, a QC laboratory, records a draft subgroup beside each official group in its 2027 plan, so the move to a revised chapter is a mapping exercise.

For LC-11, the laboratory wrote user requirements before purchase, including a column oven range of 30 to 60 °C because one planned method runs at 55 °C. The supplier’s engineer ran installation and operational tests under a protocol the laboratory approved in advance, with these results on 14 October 2026:

The laboratory could restrict LC-11’s qualified range to 30 to 45 °C, which covers every current method but needs a passing test at 45 °C, or repair it. Because the 55 °C method was due within the year, the engineer replaced the oven sensor under change control and the test was repeated: 60.0 °C read 60.9 °C, an error of +0.9 °C, a pass. Annex 15 asks OQ to include “Tests to confirm upper and lower operating limits”; here the limits came from the laboratory’s methods, not the supplier’s specification.⁸

The data system was validated as configurable software. Ongoing verification is a performance check every six months plus system suitability in every run: as Waters put it at the workshop, system suitability tests the instrument “for the method you are using with the conditions you are using”, while qualification covers the ranges the laboratory defines.¹²

5Judgement calls and common mistakes.

Practitioners at USP’s workshop named the places where laboratories go wrong, or where the draft does not yet match practice.³¹²

• One instrument, many methods. Merck noted that most laboratories run several methods on one HPLC, so “Qualification scope must cover the operational range of all methods, not just one”.
• Use outside the qualified range. Agilent’s example: “My instrument is qualified for an injection range of 10–100 µL. Can a 2 µL injection be considered acceptable?” Without a test at 2 µL, nothing shows it is.
• System suitability as PQ. Agilent reported “Labs still treating system suitability as PQ or skipping PQ entirely”. One does not stand in for the other.
• Requirements written after purchase. Merck said “The URS is often developed after instrument selection”; McDowall’s answer is a living URS, generic for selection, then updated for the instrument bought.
• Family testing. Merck described qualifying “a representative subset of identical or similar instruments within a family”. Until USP gives grouping criteria, justify each grouping in writing.
• Outsourced testing. Merck pointed out that contract testing laboratories “may use shared instruments across multiple clients”, while the sponsor “retains regulatory responsibility for data quality but lacks direct control over instrument qualification”. Ask for the qualification status of the instruments behind your results, and write it into the quality agreement.
• Split ownership. In larger companies, Merck said, metrology and validation groups handle “calibration, IQ/OQ (Qualification), and Validation (CSV)” while laboratory operations keeps the method fit for purpose. Name who signs each record.
• Repairs and moves. Lilly routes every repair through return-to-service testing, which may repeat all or a relevant subset of the original qualification tests. PIC/S lets “like for like” changes run through planned maintenance if that decision is documented, justified and approved by the quality unit.¹⁶

6What the proposed revision changes.

Whether the 4Qs survive is debated. The briefing says the 4Qs model “will be maintained although it has some weaknesses”.¹ At the workshop, Bob McDowall, an expert adviser to USP’s subcommittee, presented an integrated lifecycle that “does not use the 4Qs model”, in which design qualification becomes a selection report because “Laboratories rarely design laboratory instruments and systems”.³ Once the subcommittee has consolidated the comments, revised the draft and written a glossary, the PAL-DS Expert Committee decides whether it goes to ballot, then it is published in Pharmacopeial Forum.¹⁵ Until a revised chapter is official, keep working to IQ, OQ and PQ, which Annex 15 also uses.⁸

7What inspectors cite.

The letters below do not name <1058>; they cite what qualification is meant to prevent.

“For example, the (b)(4) (serial number (b)(4)) is the sole instrument used for endotoxin testing for both (b)(4) monitoring and final product release across all product lines. Despite being in use since December 18, 2023, this instrument has never been calibrated, and a calibration schedule has never been established.” Nipro Renal Solutions USA, warning letter, 24 July 2026¹⁹

That is a device maker cited under the Quality Management System Regulation, but its item 7 is the qualified-range mistake in plain form: the combined IOPQ “only provides testing results for products with an endotoxin specification of (b)(4)/mL”, leaving the liquid product lines and the retest procedure uncovered.¹⁹

“Your GC laboratory instrument was not adequately qualified because you failed to validate the control software or enable audit trails, and the GC data file creation and modification dates were not accurate.” Laboratorios Jaloma, warning letter 320-26-78, 22 May 2026²⁰

FDA treated unvalidated control software and missing audit trails as a failure of qualification. In August 2026 it also cited K.C. Pharmaceuticals because it “routinely performed total organic carbon (TOC) and conductivity measurements without prior system suitability of the analyzer”:²¹ qualification and day-of-use checks are both needed.

8How it runs in Seal.

In Seal, equipment records connect qualification, calibration, cleaning and maintenance evidence to the workflow that selects each instrument. A release test can require a qualified range, a current calibration and an authorised analyst, and a failed or missing prerequisite stops that path with the reason recorded. Recorded uses link to samples and batches, so when a check fails, such as a failed flow check on an HPLC, the results that relied on it since the last acceptable check come from the record rather than being rebuilt by hand. Calibration records keep each point’s as-found and as-left values and the reference standard used, with criticality taken from intended use.

For new equipment, Seal’s CQV records link each requirement to its risks, tests, exceptions and release decision, and supplier evidence such as factory acceptance results earns credit only under an approved strategy. Seal can also collect the original file from any instrument software that saves one, linked to its sample, method and run.