Good Laboratory Practice (GLP) interview questions

What a strong answer covers

• GLP is a quality system for the planning, performance, monitoring, recording, reporting and archiving of non-clinical safety studies, the studies that assess the safety of a substance before or alongside its use in humans, animals or the environment.
• Its purpose is reconstructability and mutual confidence: a regulator reviewing the report years later should be able to trace every reported number back to the raw data and to the person and procedure that produced it.
• GMP governs manufacturing quality of products, and GCP governs the conduct of clinical trials in human subjects. They share a documentation and data integrity philosophy but apply to different activities and are assessed against different requirements.
• GLP is not a measure of scientific merit or of how good the science is. A study can be fully GLP compliant and scientifically weak, which is a distinction interviewers like to hear made.
• A strong answer names the framework it is speaking about, for example the OECD Principles of GLP or the relevant national regulation for the country the facility operates in, and notes that facilities are monitored by a national compliance authority whose identity varies by country.
• Not every laboratory activity in a company is under GLP. Discovery and research work often is not, and knowing which of your work is in scope is part of working in a mixed facility.

Where people lose the point

• ×Describing GLP as being about doing good science, rather than about verifiable, reconstructable study conduct.
• ×Conflating GLP with GMP or GCP, or claiming a single global rulebook applies everywhere.
• ×Assuming all laboratory work in a regulated company is GLP work.

What a strong answer covers

• An SOP is an approved, version-controlled procedure describing how a routine activity is performed, so that the activity is done the same way regardless of who does it, and so a reader can later know exactly how it was done.
• Working to SOPs means having the current approved version available at the point of work, being trained on it with that training documented, and not working to a personal or remembered version.
• If an SOP cannot be followed as written, you stop and escalate rather than improvising. Deviating silently converts a procedural problem into a data integrity problem.
• If the departure has already happened, it is recorded as a deviation with its cause and an assessment of its impact, and the study director is informed because they carry responsibility for study integrity.
• If the SOP is simply wrong or unworkable, the fix is to revise it through change control, not to keep deviating from it repeatedly. A recurring deviation against the same SOP is itself a finding.
• A strong answer mentions that superseded versions are withdrawn from use and retained in the archive, so it is always possible to know which version was in force on a given date.

Where people lose the point

• ×Improvising a workaround and telling nobody because the result looked fine.
• ×Treating repeated deviations as normal rather than as a signal the SOP needs revising.
• ×Working from an uncontrolled copy, a printout of unknown vintage or personal notes.

What a strong answer covers

• Raw data is the original record of an observation or measurement, and everything reported in the study report has to be traceable back to it. That includes lab notebook entries, instrument output, worksheets, photographs and the electronic files an instrument produced.
• For a computerised system, the original record is the electronic data together with its metadata and audit trail, not a printout. Keeping only the paper copy generally loses information that was part of the original.
• Apply ALCOA+ as a test: attributable, legible, contemporaneous, original, accurate, and also complete, consistent, enduring and available.
• Corrections are made by a single line through the error leaving it readable, the corrected entry, and the initials, date and reason of the person making the change. Obliteration, overwriting and correction fluid are not acceptable.
• Contemporaneous means recorded as the work happens. Writing up at the end of the day from memory is a common real-world failure and it is the one that is hardest to defend in an inspection.
• A strong answer mentions that a copy can stand in for the original only when it is a verified true copy made under a defined procedure, and that facility procedures define how and when that is allowed.

Where people lose the point

• ×Calling a transcribed summary or a printed report the raw data.
• ×Discarding electronic source files once a printout exists.
• ×Backfilling entries or pre-signing forms, which destroys the contemporaneous character of the record.

What a strong answer covers

• The study director is the single point of study control: one named individual with overall responsibility for the technical conduct of the study and for the interpretation, analysis, documentation and reporting of results.
• They approve the study plan and any amendments, ensure study personnel have access to it, and ensure the study is conducted according to the plan, the applicable SOPs and the GLP framework.
• They are responsible for ensuring that deviations are recorded and assessed for impact, and that unforeseen circumstances affecting quality and integrity are dealt with rather than absorbed.
• At the end, they sign the report, and that signature is the statement that the study was conducted in compliance and that the report describes it accurately.
• They ensure that the study plan, raw data, samples, specimens and the final report are transferred to the archive.
• A strong answer notes the difference between the study director's role and that of a principal investigator at a test site in a multi-site study, and that test facility management, not the study director, is responsible for appointing the study director and providing the resources the study needs.

Where people lose the point

• ×Describing the role as purely administrative sign-off rather than accountability for technical conduct and integrity.
• ×Sharing the responsibility across a committee, when the framework deliberately places it with one named person.
• ×Confusing the study director's responsibilities with those of test facility management or of the quality assurance unit.

What a strong answer covers

• Independence exists so that the people assessing compliance have no stake in the study's outcome or schedule. Personnel performing quality assurance must not be involved in conducting the study they are assessing.
• QA inspects: study-based inspections at critical phases, facility-based inspections of general operations, and process-based inspections of repeated procedures. It also reviews study plans and reports against the raw data.
• QA maintains copies of study plans and SOPs, keeps records of inspections, and reports findings to management and the study director promptly enough that problems can be corrected while the study is running.
• The QA statement in the final report records what was inspected and when it was reported, which is how a reader knows the report was independently checked.
• QA is not there to do the science or to fix the study. It reports findings; correction sits with the study director and management.
• A strong answer frames QA as an ally rather than an adversary: an inspection finding raised during the study is far cheaper than the same problem found by a regulator afterwards.

Where people lose the point

• ×Placing QA under the same manager as study conduct and calling it independent.
• ×Describing QA as a proofreading service for reports.
• ×Treating findings as personal criticism instead of as inputs to correction and preventive action.

What a strong answer covers

• The system should be validated for its intended use, with documented evidence that it does what it is supposed to do in the environment it actually runs in, and it should be kept under change control afterwards.
• Access must be controlled with individual accounts and appropriate privileges, because a shared login destroys attributability. Privileges to alter or delete data should be limited and separated from routine users.
• A secure, computer-generated, time-stamped audit trail should record creation, modification and deletion of records, including who did it and why where a reason is required, and it must not be switchable off by ordinary users.
• Audit trails only work if they are reviewed. A facility should define who reviews them, how often and against what, because an unreviewed audit trail catches nothing.
• Backup, restore and archiving must be defined and tested, and the data must remain readable for the retention period, which is a real problem for legacy instrument formats.
• A strong answer mentions that system clocks need to be synchronised and controlled, since timestamps are the backbone of the contemporaneous record, and that electronic signature requirements depend on the applicable regulation for that market.

Where people lose the point

• ×Accepting shared logins for convenience on a shared instrument.
• ×Enabling an audit trail and never reviewing it.
• ×Treating validation as a one-off exercise rather than something maintained through change control.

What a strong answer covers

• You raise it. Promptly and in writing, to the study director in the first instance, and through quality assurance and management according to the facility's procedures. A signed report does not close the door, and the timing is what tests integrity.
• Establish the facts before characterising them: what the raw data actually says, what the report says, whether the difference is a transcription error, a calculation error, or something that changes a conclusion.
• Assess impact. A typographical error in a table footnote and an error that changes a no-observed-effect determination are handled with the same honesty but with very different urgency and reach.
• The correction route is a formal amendment to the final report, issued under the same controls as the report itself, describing the change and the reason for it. The original report is not quietly edited or replaced, and the study file must show what changed and why.
• If the report has already gone to a sponsor or a regulator, the facility's procedures and the applicable regulatory obligations determine who must be notified, and that decision is management's rather than an individual's to make alone.
• A strong answer adds a cause investigation: how did the transcription or checking step fail, and what changes to procedure or review prevent recurrence.
• It is also worth saying what you would do if you were discouraged from raising it: escalate through quality assurance or the facility's defined route, and keep your own contemporaneous record of what you reported and when.

Where people lose the point

• ×Deciding it is too small to mention, or waiting to see whether anyone notices.
• ×Correcting the issued report in place with no amendment record and no trace of the original.
• ×Letting commercial pressure or the awkwardness of the timing settle the question.

What a strong answer covers

• At study completion the study plan, raw data, samples and specimens where applicable, and the final report are transferred to the archive under the control of a named archivist.
• Access is controlled and recorded: entries and removals are logged, so the chain of custody of a record is known throughout its life.
• Conditions must preserve the records for their retention period, which covers physical conditions for paper and specimens and media and format viability for electronic data.
• Retention periods are set by the applicable regulation, the sponsor's requirements and the facility's procedures, and they differ between jurisdictions and study types, so the honest answer is that you follow the defined period for that study rather than quoting a single number.
• Retrievability is the hard part for electronic records: the data must remain readable, which can mean migrating formats or maintaining the ability to read legacy instrument files, and any migration itself has to be controlled and verified.
• A strong answer mentions that arrangements must survive the facility itself, so if a facility closes, the archive has to be transferred to another party under defined arrangements rather than abandoned.

Where people lose the point

• ×Quoting a single universal retention period as though it applied everywhere.
• ×Archiving electronic data on media or in a proprietary format that will not be readable for the retention period.
• ×Leaving the archive as an unlocked storeroom with no access log and no named responsible person.

What a strong answer covers

• Decline, and be plain about why: a signature is a factual claim that you did or witnessed something, and signing for work you did not observe makes the record untrue, which is a data integrity issue rather than a paperwork preference.
• Separate the person from the problem. The pressure is usually about the schedule, so offer the route that is both legitimate and fast: find the person who did the work to sign, or record the position honestly and raise the timing problem to the study director.
• If the request persists, escalate through the defined route, which typically means quality assurance or management, and record what was asked and what you did.
• Keep your own contemporaneous note. If the matter is ever revisited, the difference between remembering and having recorded it at the time is the whole difference.
• A strong answer treats this as a systems question too: repeated pressure of this kind usually indicates unrealistic scheduling or under-resourcing, and that is worth raising as a cause rather than as a complaint about one incident.
• It also acknowledges the human reality without conceding: you can be respectful, assume good faith and still refuse, and a good facility will back you for refusing.

Where people lose the point

• ×Signing to keep the peace and intending to mention it later.
• ×Refusing in a way that makes it a confrontation rather than a procedural point, which makes escalation harder.
• ×Escalating without any record of what was asked, when and by whom.

What a strong answer covers

• The honest first answer is that readiness is a state, not a project: a facility that works to its procedures every day has much less to do, and a scramble in the final month is itself a finding waiting to happen.
• Practical preparation includes confirming that training records, SOP versions, equipment maintenance and calibration records and archive indexes are current and retrievable, since inspectors frequently sample exactly these.
• Review open deviations, findings and corrective actions: an open item with a clear plan and evidence of progress is defensible, while one that has sat untouched since it was raised is not.
• Prepare people as well as documents. Staff should be able to describe what they do and find the SOP that governs it, and should understand that answering the question asked, honestly and without speculating, is what is wanted.
• Set up the logistics: a room, a document runner, a scribe recording every request and response, and a route for questions the responder cannot answer immediately.
• A strong answer says explicitly that nothing is fixed retrospectively during preparation. Finding a gap is legitimate; back-dating a record to close it is fraud.

Where people lose the point

• ×Treating preparation as tidying, and manufacturing records to fill gaps found during the review.
• ×Coaching staff to volunteer information beyond the question, or to guess when they do not know.
• ×Failing to keep a log of what documents were provided during the inspection.