Batch record review is the moment a company decides, on paper, whether a batch is fit to release. Everything upstream of that review — the manufacturing, the testing, the sampling — is preparation. The review itself is the decision point, and it's also the single most examined activity in almost every FDA inspection and third-party audit I've been part of. Investigators don't just check whether a record exists. They check whether the review that approved it actually did its job.
That distinction matters more than most quality teams treat it. A signature at the bottom of a batch record isn't compliance. It's a claim. And auditors have gotten very good at testing whether the claim holds up.
What GMP Actually Requires for Batch Record Review
The core U.S. requirement sits in 21 CFR 211.192, which states that any unexplained discrepancy — including a failure of a batch to meet its specifications — must be investigated, whether or not the batch has already been distributed. That single clause is doing a lot of work. It means the reviewer's job isn't to confirm paperwork is complete. It's to notice what's missing, what's inconsistent, and what doesn't add up, and to chase it down before the batch moves forward.
21 CFR 211.188 sets the baseline: batch production and control records must be prepared for each batch and must include a complete history of the production of that batch. 211.194 extends the same logic to laboratory records — test methods, results, and the identity of the analyst and reviewer all have to be documented and checked. Together, 188, 192, and 194 form the backbone of batch documentation in U.S. drug GMP, and in my experience they're also the three sections auditors cite most often when a batch record review program has gaps.
ICH Q7, which governs active pharmaceutical ingredient manufacturing, states the review requirement even more directly in section 2.20: batch production and control records should be reviewed and approved by the quality unit before a batch is released or distributed. The EU takes the same position through EudraLex Volume 4, Chapter 4, and Annex 11 adds a specific expectation for electronic systems — audit trails covering data creation, modification, and deletion have to be reviewed, not just retained.
None of these regulations describe how to review a batch record. That's left to the manufacturer, and it's exactly where I see the most variability between companies that sail through inspections and companies that don't.
Who Is Required to Review a Batch Record
GMP requires an independent quality unit review, separate from the personnel who performed the manufacturing or testing. This isn't a formality. A production supervisor reviewing their own department's record has an incentive, even an unconscious one, to read past a gap rather than stop and question it. Independent review breaks that incentive. Most mature quality systems layer two levels: a first-pass production or manufacturing review that checks completeness, and a second, independent quality assurance review that has release authority. FDA's expectation, backed by 211.192, is that the second reviewer is empowered to hold the batch, not just to co-sign it.
The Two Ways Companies Actually Review Batch Records
In practice, I see two dominant approaches, and they produce very different audit outcomes.
Line-by-line review treats every entry, every initial, every calculation as something to verify against the master batch record. It's thorough and it's defensible, but it's slow, and reviewer fatigue sets in on long records. I have watched experienced reviewers miss an obvious out-of-sequence entry on page 40 of a 60-page record simply because attention degrades over that many identical checks.
Risk-based review concentrates reviewer attention on the steps most likely to affect product quality — critical process parameters, in-process test results, yield calculations — while giving routine entries a lighter check. Done well, this approach actually catches more real problems, because the reviewer isn't burned out by the time they reach the section that matters. Done poorly, it becomes an excuse to skip sections nobody wants to deal with, and that's exactly what an auditor will probe for.
| Approach | Where it succeeds | Where it fails | What auditors probe |
|---|---|---|---|
| Line-by-line review | Complete records, straightforward process | Reviewer fatigue on long batches; slow release cycle | Whether checks were genuinely performed or rubber-stamped |
| Risk-based review | Catches critical deviations faster; scales with volume | Can mask gaps if criticality wasn't formally assessed | Whether the risk assessment behind the approach is documented and approved |
| Real-time (concurrent) review | Deviations caught while the batch is still on the floor | Requires staffing on production shifts | Whether concurrent review actually happened, or was backfilled after the fact |
| Post-production review only | Simple to staff and schedule | Deviations discovered days after the fact, when memory and context have faded | Time gap between manufacture date and review completion date |
I have come to think the real dividing line isn't line-by-line versus risk-based. It's whether the review happens close to the manufacturing event or gets batched up and done later, under release-date pressure. A record reviewed the same day the batch was made gets read carefully. A record reviewed three weeks later, with a dozen other batches stacked behind it, gets read quickly.
What Auditors Actually Flag
This is the part quality teams most want to know, and it's worth being specific rather than general. In my work reviewing client batch records before FDA and notified body audits, the same categories come up again and again.
Blank fields and missing initials
This sounds minor. It isn't. A blank field on a batch record raises an immediate question: was the step performed and not documented, or was it not performed at all? Under 211.192, that ambiguity is itself a discrepancy requiring investigation. Auditors will pull a blank field and ask the reviewer, on the spot, how they closed it out. "We assumed it was fine" is the answer that turns a documentation observation into a data integrity finding.
Corrections that don't follow ALCOA+ principles
A crossed-out entry with no date, no initials, and no reason for the change is one of the most commonly cited findings in FDA Form 483s and warning letters involving batch documentation. The correction itself usually isn't the problem — GMP allows corrections. What draws the citation is the missing context: who made the change, when, and why. Good corrections are legible, dated, initialed, and explained. Anything less reads as an attempt to obscure rather than clarify, whether or not that was the intent.
Deviations documented in the batch record but not linked to a formal investigation
I see this constantly. An operator notes "temperature excursion, resolved" directly in the batch record margin, and the note never generates a deviation report, a root cause investigation, or a CAPA. The batch gets released anyway. This is one of the fastest paths to a serious observation, because it demonstrates that the quality system isn't actually functioning as a system — deviations are being managed informally at the point of manufacture instead of through the process 211.192 requires.
Yield and reconciliation discrepancies that go unexplained
Theoretical yield versus actual yield is one of the first calculations any competent auditor checks. A yield that falls outside the expected range without a documented explanation is an automatic flag, and reviewers who sign off on unexplained yield variances are, in effect, signing off on an unresolved discrepancy.
Review completed after the batch has already shipped
Timing gets checked. If the batch record shows a manufacture date, a QA release signature date, and a distribution record, and the release signature postdates distribution, that is a serious finding regardless of how clean the rest of the record looks. It suggests the review is a formality performed after the decision was already made on the floor.
Audit trail review gaps in electronic batch records
As more manufacturers move to electronic batch record systems, this has become one of the fastest-growing categories of findings. Annex 11 and FDA's data integrity expectations both require that audit trails — the system-generated log of who changed what and when — get reviewed as part of batch disposition, not just stored in case someone asks. A company that has an audit trail feature turned on but never actually looks at it is, in an inspector's eyes, no better off than a company using paper.
The pattern across all of these is the same. GMP doesn't just require documentation. It requires that documentation be read, questioned, and reconciled by someone with the authority to stop the batch. Auditors aren't looking for perfect records. They're looking for evidence that imperfect records got caught.
Building a Batch Record Review Program That Holds Up
A few practices consistently separate the companies that pass cleanly from the ones that collect observations.
Train reviewers on what to look for, not just how to sign. Most batch record review training I've audited focuses on the mechanics of where to initial and which boxes to check. Very little of it trains reviewers to ask "does this make sense" when they hit an entry that's technically complete but substantively odd. That judgment is exactly what 211.192 is asking for, and it has to be taught, not assumed.
Set a real time limit between manufacture and review. Companies that require review within 24 to 48 hours of batch completion catch far more genuine problems than companies that let records queue for a week. The gap between event and review is where memory, context, and the ability to actually resolve a discrepancy all degrade.
Separate first-pass completeness checks from second-pass quality disposition. A production reviewer confirming the record is complete is doing something different from a QA reviewer deciding whether the batch is fit to release. Blurring those two roles into one signature is a common shortcut, and it's one auditors specifically probe for by asking each signer to explain, separately, what they were checking.
Treat the electronic batch record's audit trail as part of the record, not an add-on. If your system generates an audit trail, your review procedure needs to say, in writing, who reviews it and how often. An unreviewed audit trail is worse than no audit trail at all, because it demonstrates the capability existed and wasn't used.
Escalate ambiguity instead of resolving it silently. The habit that protects a quality system best is simple: when a reviewer isn't sure whether something is fine, the default should be to ask, not to assume. That single habit, applied consistently, prevents more findings than any procedure rewrite.
None of this is exotic. It's disciplined follow-through on requirements that have existed in some form since Part 211 was written. The companies that struggle aren't usually missing the regulation. They're missing the follow-through.
If you want an outside read on where your batch record review process actually stands against 21 CFR 211.192 and ICH Q7 expectations, a data integrity gap assessment is usually the fastest way to find out before an inspector does. And if deviations are surfacing in your batch records without a clean path to CAPA, our overview of deviation and CAPA management walks through how to close that loop.
Frequently Asked Questions
What regulation requires batch record review under GMP? 21 CFR 211.192 requires investigation of any unexplained discrepancy in a batch, and 211.188 and 211.194 require complete production and laboratory records. ICH Q7 section 2.20 requires quality unit review and approval before release. EU manufacturers follow the equivalent requirement in EudraLex Volume 4, Chapter 4.
Who is qualified to review a batch record? GMP expects an independent quality unit reviewer, separate from the personnel who performed the manufacturing or testing, with the authority to hold or reject the batch. Many companies layer a first production-level completeness check under a second, independent QA disposition review.
What is the most common batch record finding in FDA inspections? Unexplained discrepancies and incomplete investigations tied to 21 CFR 211.192 are among the most frequently cited issues in drug manufacturing Form 483s, alongside corrections that lack a date, initials, or documented reason.
Do electronic batch records eliminate review findings? No. Electronic batch records reduce transcription errors and improve traceability, but they introduce a new expectation: the system's audit trail has to actually be reviewed as part of batch disposition, not just retained. An unreviewed audit trail is itself a common audit finding.
How long should a company wait between manufacturing a batch and reviewing the record? There's no fixed regulatory number, but review completed within 24 to 48 hours of batch completion consistently catches more genuine discrepancies than review performed after a backlog builds, because the people involved still remember the context needed to resolve a question.
Last updated: 2026-08-07
Jared Clark
GMP Compliance Consultant, Certify Consulting
Jared Clark is a GMP compliance consultant and founder of Certify Consulting, specializing in FDA GMP requirements for pharmaceuticals, dietary supplements, cosmetics, and food manufacturing.