Good Laboratory Practice (GLP) Basics: A 21 CFR Part 58 Overview

Spend any time with laboratory paperwork or supplier documentation and you've seen the letters GLP. Study reports, vendor marketing, regulatory filings — they turn up everywhere, and almost nobody explains them. That gap matters for anyone sourcing compounds labeled for research use only, because GLP gets invoked as though it were a quality stamp on a product. It isn't one.

Good Laboratory Practice under 21 CFR Part 58 is the FDA regulation governing how nonclinical safety studies are planned, run, recorded, reported, and archived. It's a rule about process and paperwork — not scientific merit, not product grade. Below: what it covers, what it excludes, who is accountable, how FDA enforces it, and why "GLP grade" is not a term the regulation recognizes.

What 21 CFR Part 58 Actually Is

Part 58 is the FDA regulation titled "Good Laboratory Practice for Nonclinical Laboratory Studies," and FDA describes its requirements as intended to ensure the quality and integrity of safety data obtained from animal studies submitted to FDA. In the broader literature, GLP is defined as a quality program related to organizational processes and conditions where non-clinical health and environmental safety studies are planned, performed, monitored, recorded, reported and archived. Look at what that list contains: six verbs, every one of them procedural.

The clearest way to hold it is this. Part 58 doesn't grade a study's science. It takes no position on whether a finding is biologically meaningful or whether a sponsor's conclusion is defensible. What it governs is whether the study can be reconstructed and checked. A GLP study can be scientifically mediocre and fully compliant at the same time; a brilliant study with sloppy records can fail outright.

Why Part 58 Exists: The Industrial Bio-Test Collapse

Regulations like this one rarely come from abstract principle. Part 58 exists because a contract laboratory got caught inventing data on an industrial scale. Industrial Bio-Test Laboratories, of Northbrook, Illinois, had submitted thousands of safety tests to regulators that were either falsely claimed to have been performed or were of such poor quality that investigators could not determine the extent of actual work completed. The affected submissions covered drugs, insecticides, herbicides, food additives, pesticides, cosmetics, and cleaning products.

The peer-reviewed account of the period is specific about the failure modes: inadequate study planning, insufficient documentation, fraudulent data including replacement of deceased animals without proper treatment records, and manipulation of raw data to "adjust results" for final reports. Each of those has a matching requirement in Part 58 today. The regulation reads less like a philosophy of good science and more like a list of things that went wrong once.

Congressional hearings in the mid-1970s produced an FDA proposed rule on 19 November 1976, a final rule effective 20 June 1979, and amendments effective 5 October 1987. EPA built parallel frameworks for pesticides and industrial chemicals at 40 CFR 160 and 40 CFR 792.

What Counts as a Nonclinical Laboratory Study

The short answer: an experiment in which test articles are studied prospectively in test systems — animals, plants, microorganisms, or subparts of them — under laboratory conditions, to determine their safety. Studies in human subjects sit entirely outside Part 58.

What pulls a study into scope is intent. Part 58 attaches to studies conducted to support an application to FDA for a research or marketing permit; subject matter alone doesn't do it. That logic will feel familiar to anyone who has read the intended-use doctrine, where what a party means to do with something, rather than what the label says, drives regulatory status.

FDA's long-standing guidance is unusually direct about the exclusions. Falling outside Part 58: validation trials for analytical methods, cosmetic safety studies, organoleptic evaluation of processed foods, and field trials in animals. Basic research and exploratory studies run with no plan to submit the data are outside the rule too, though FDA notes the regulations still describe quality standards laboratories should pursue.

The same guidance draws a sharp, slightly counterintuitive line through analytical work. Chemistry that characterizes the test article, assesses its stability, or determines the homogeneity and concentration of a test mixture is covered when it feeds a covered study. Developing the analytical method, and validating that it works, is not. Scope also follows the work out to vendors: subcontracted pathology reads, slide preparation, and statistical analysis are covered "to the extent that they contribute to a study" subject to the regulation. A sponsor can't shed the obligation by moving a piece of the study into someone else's building.

The Three Accountabilities: Management, Study Director, QAU

Part 58 splits responsibility deliberately, so that nobody both produces data and certifies it. Three roles carry it.

Testing-facility management

Management answers for the facility as an institution: adequate resources, qualified personnel, written standard operating procedures, and an overall posture of compliance. In the OECD framing of the same structure, management bears ultimate responsibility for organizational functioning and quality program success.

The Study Director

The Study Director is the single point of scientific and procedural control for one specific study. Per FDA's own primer on the regulation, that means approving the protocol, keeping conduct aligned to it, reviewing raw data, signing the final report, and holding authority over study personnel. A deputy may step in during an absence — but only if designated and documented.

Two constraints matter. A Study Director stays responsible for protocol deviations discovered after a study closes; those go into the final report rather than quietly disappearing. And the same person can't serve as both Study Director and Quality Assurance Unit, nor hold the chief-executive role at the same time.

The Quality Assurance Unit

The QAU must be entirely separate from and independent of the personnel directing and conducting the study, and its head cannot report to the Study Director. Its work is procedural: maintain current protocols as amended and the relevant SOPs, inspect studies periodically, review amendments to final reports, confirm that computer-formatted data faithfully represents the underlying raw data, and verify that the final report reflects what the raw data shows.

What the QAU doesn't do is redo the science. FDA's guidance says plainly that the unit "is not expected to perform a scientific evaluation" or to substitute its judgment for that of the subject-matter experts. The separation is the whole point. Industrial Bio-Test failed in part because the people generating the data were the only people checking it.

Characterizing the Test Article

A safety study is only as good as its knowledge of what entered the test system. So Part 58 requires that test and control articles be characterized for identity, strength, purity, and stability — and where an article is mixed into a carrier, for the homogeneity and concentration of that mixture. That characterization chemistry falls inside GLP scope when it contributes to a covered study.

Anyone who has worked through a supplier's analytical paperwork will recognize the questions, because they're the same ones: is this the compound it claims to be, how pure is it, how much water is in the powder, what counterion came along with it. Our guide to reading a certificate of analysis field by field covers how those answers get presented for an individual lot. Keep one distinction sharp, though. A CoA answers the characterization question for a vial; GLP answers it for a regulated study. A CoA is not a GLP record, and Part 58 doesn't require one.

Raw Data, Archives, and the Reproducibility Standard

FDA defines raw data as "any laboratory worksheets, records, memorandum, notes that are the result of original observations." That sweeps in more than most people expect — quarantine records, animal receipt documentation, environmental monitoring logs, and equipment calibration records all qualify. Cage cards count only when they carry original observations, not routine identifiers. Photocopies that are dated and verified by the person who made them are acceptable as exact copies, and raw data need not live in bound notebooks so long as it is identified and retrievable.

Specimens are kept for the regulatory term "or for as long as their quality permits meaningful reevaluation, whichever is shorter" — a concession that no retention schedule can outrun the chemistry of a degrading sample. Records go into secure storage protected against deterioration and unauthorized access.

Underneath all the filing requirements sits one standard. A third party — potentially years later, with none of the original personnel available — should be able to reconstruct the full course of the study from the archive: what the test systems received, what was collected, how the results were compiled. GLP requires methods transparent and detailed enough that a different laboratory following identical procedures reaches the same result. That's what makes the regulation workable. It lets a regulator judge data integrity by reconstruction rather than by repeating expensive and ethically fraught animal work.

How FDA Enforces Part 58

GLP inspections run through FDA's Bioresearch Monitoring program, known as BIMO, which also carries 21 CFR 812, 50, and 56. Inspections come in two modes: routine surveillance assignments covering facilities selected under the compliance program, and directed — "for cause" — inspections triggered by a submission review, a data problem, or a complaint.

FDA's internal compliance program for these inspections sets out three questions. Was the testing performed according to the submitted protocol? Would any reported deviations have affected the reliability of the test results? Does the final report accurately reflect the raw data? An inspection may include a detailed data audit of one or more specific studies; investigators present credentials and issue Form FDA-482 at the outset.

Every inspection closes with a classification: NAI (No Action Indicated) where the establishment is in compliance, VAI (Voluntary Action Indicated) for minor deviations needing correction, and OAI (Official Action Indicated) for significant failures. Observed deviations are listed on Form FDA-483. From there the escalation is graded — untitled letter, warning letter, rejection of the study data from the application it supported, and at the far end, disqualification of the facility. Disqualification bars it from generating data for FDA submissions at all, until it petitions for reinstatement with documented corrective action behind it.

GLP Is Not GMP, and Not GCP

Three quality systems get conflated constantly, and that conflation is where most of the confusion about "GLP" starts. GLP governs nonclinical safety studies. GMP governs manufacturing. GCP governs studies in human subjects. Different questions, and none of them stands in for another.

The consequence for sourcing is specific: a compound synthesized entirely outside any GMP system can legitimately serve as the test article in a GLP study. Part 58's requirement is that the study laboratory characterize what it has — not that a particular kind of facility produced it. Where the manufacturing line actually falls is a separate question, and we covered it in our piece on GMP and research-grade manufacturing.

Which brings us to the phrase on the product page. Part 58 doesn't define a product grade, so "GLP grade" is not a term the regulation recognizes. A supplier can accurately state that its material has been used as a test article in GLP studies. It cannot make GLP a property of the vial. The adjacent labeling question is the subject of what research-use-only labeling actually requires.

GLP as an International Standard

The OECD adopted GLP into its Test Guidelines in 1981 and revised the Principles in 1997, producing a harmonized standard covering chemical testing in laboratories, greenhouses, and field experiments — while explicitly excluding studies in human subjects. The payoff is the Mutual Acceptance of Data system, under which a study run to GLP in one adhering country is accepted for assessment in the others. Adherence reaches beyond OECD membership to countries including Brazil, India, Malaysia, Singapore, and South Africa. GLP is, in short, the common language that keeps the same toxicology study from being run again in every market.

Frequently Asked Questions

Is 21 CFR Part 58 the same thing as the OECD Principles of GLP?

No, though they cover the same ground and were deliberately harmonized. 21 CFR Part 58 is United States law, enforceable against testing facilities whose data supports an FDA submission. The OECD Principles are an international standard that adhering countries implement through their own national rules — the European Union, for instance, through Directives 2004/9/EC and 2004/10/EC.

Does a research chemical have to be made in a GLP facility?

No — and this is one of the most common mix-ups. GLP governs how a safety study is run, not how a chemical is manufactured. Part 58 applies to the laboratory conducting the study, not to the supplier who synthesized the test article. A compound produced entirely outside any GLP or GMP system can legitimately serve as a test article, provided the study laboratory characterizes it as the regulation requires.

Do early exploratory laboratory studies have to follow GLP?

Generally not. Part 58 reaches studies intended to support an application to FDA for a research or marketing permit. Early screening work — initial exposure-level range finding, preliminary genotoxicity screens, early pharmacokinetic characterization — is usually run outside GLP, deliberately, because smaller studies move faster and cost less. FDA notes the framework still describes quality practices such laboratories should aspire to.

What is the difference between an FDA Form 483 and facility disqualification?

A Form FDA-483 is the list of inspectional observations an investigator leaves at the end of an inspection, which the facility is expected to address. Disqualification sits at the other end of the scale: FDA formally bars the facility from generating data for FDA submissions. Between the two sit untitled letters, warning letters, and rejection of the study data. A disqualified facility must petition for reinstatement with documented corrective action.

Who is ultimately responsible if a GLP study goes wrong?

Part 58 splits accountability three ways so no single person can both run a study and certify it. Testing-facility management answers for resources, personnel, and the facility's compliance posture. The Study Director holds single-point control over one study — protocol, conduct, raw-data review, signed final report. The Quality Assurance Unit audits independently and cannot report to the Study Director.

What to Take From Part 58

Part 58 is a documentation regulation born from documented fraud. It doesn't tell you a study's conclusions were right. It tells you the study can be reconstructed, audited, and checked by someone who wasn't there — which, after Industrial Bio-Test, was the problem worth solving.

The useful consequence is a better question. When "GLP" shows up on paperwork, the thing to ask isn't "is this good material." It's: whose study, and is that study inside Part 58's scope at all? If the answer names a specific study at a specific facility submitted to FDA, the phrase means something concrete. Printed on a vial, it means nothing.

For research use only. Not for human or animal consumption of any kind. The information in this article is for educational purposes only and is not intended to diagnose, treat, cure, or prevent any disease. The statements made have not been evaluated by the U.S. Food and Drug Administration. These products are NOT FDA APPROVED. Please consult with a licensed healthcare professional before making any decisions regarding your health or research.

Optides LLC is a chemical supplier. Optides LLC is not a compounding pharmacy or chemical compounding facility as defined under 503A of the Federal Food, Drug, and Cosmetic Act. Optides LLC is not an outsourcing facility as defined under 503B of the Federal Food, Drug, and Cosmetic Act.