21 CFR Part 58: A Guide to Good Laboratory Practice (GLP)

21 CFR Part 58, titled Good Laboratory Practice for Nonclinical Laboratory Studies (GLP), establishes quality standards for covered nonclinical safety studies that support, or are intended to support, FDA research or marketing permit applications. Issued in 1978 and codified in Title 21 of the Code of Federal Regulations, Part 58 sets requirements for the organization, personnel, facilities, equipment, protocols, and records used in those studies (^[1]). The principal goal is to ensure that safety study data are reliable, reproducible, and auditable, thereby protecting public health through trustworthy safety evidence. Unlike technical validity of a scientific hypothesis, GLP focuses on the process of data collection: requiring thorough record-keeping, independent quality assurance oversight, and controlled procedures. In essence, GLP is “less about what you found and more about proving how you found it – cleanly, consistently, and under independent QA oversight” (^[2]).

GLP’s origins trace to the late 1970s when federal investigations uncovered widespread misconduct and data fraud in private toxicology labs. The infamous Industrial Bio-Test Laboratories case, among others, prompted Congress and OSHA to empower FDA (and later EPA) to regulate lab practices. The resulting GLP rules (FDA’s in 1978 (^[3]) and EPA’s around the same time) mandated written protocols, calibrations, and comprehensive archiving. Over time, GLP has become globally harmonized through OECD principles so that data from any OECD-member lab following GLP are mutually accepted under the Mutual Acceptance of Data (MAD) system (^[4]) (^[5]). Today, thousands of nonclinical studies worldwide adhere to GLP standards.

This report provides an in-depth analysis of 21 CFR Part 58, covering its history, scope, structure (subparts and key requirements), interaction with other regulatory frameworks (EPA, OECD, EU), implementation in practice (including inspection and enforcement), data integrity and quality assurance mechanisms, and future directions. It incorporates multiple perspectives — from regulators, industry, academia, and public health — and includes case examples of GLP compliance and noncompliance. The report also examines data and evidence on GLP’s impact on trust in regulatory science, including both supportive and critical viewpoints from the literature (e.g., proponents emphasizing GLP’s role in standardization (^[6]) vs. critiques highlighting that GLP does not guarantee scientific validity (^[7])). Finally, it discusses emerging challenges such as electronic recordkeeping (21 CFR Part 11), global collaboration, and adaptive changes in the era of big data and advanced toxicology. Throughout, the discussion draws on regulatory texts, FDA guidance, OECD documentation, and peer-reviewed studies; readers should consult the cited primary sources for the requirements applicable to a particular study or submission.

21 CFR Part 58 implements the Good Laboratory Practice (GLP) regulations for nonclinical safety studies in the United States under authority of the Federal Food, Drug, and Cosmetic Act (FD&C Act) and Public Health Service Act. The FDA explicitly states that Part 58 “prescribes good laboratory practices for conducting nonclinical laboratory studies that support or are intended to support applications for research or marketing permits for products regulated by the FDA” (^[3]). This includes studies on food additives, color additives, drugs (human and animal), devices, biologics, and more (see (^[3])). Part 58 does not cover studies using human subjects, clinical studies, field trials in animals, or basic exploratory studies. It applies to prospective nonclinical laboratory safety studies of regulated test articles that support, or are intended to support, FDA research or marketing permit applications. The intent is “to assure the quality and integrity of the safety data” that underlie FDA decisions (^[3]). In simple terms, GLP is a quality system for lab studies: it sets requirements for how studies are planned, performed, monitored, recorded, and reported, with an emphasis on traceability and credibility of the data (^[5]).

The OECD Principles of GLP (adopted by the U.S. and other countries) define GLP as a “managerial quality control system covering the organizational process and the conditions under which non-clinical health and environmental studies are planned, performed, monitored, recorded, reported and retained (or archived)” (^[5]). In practice, compliance means having written Standard Operating Procedures (SOPs), trained and qualified staff, proper facilities and equipment, detailed protocols, a Study Director overseeing each study, an independent Quality Assurance Unit (QAU), and complete records of raw data and study reports (^[3]) (^[8]). GLP is often described as a quality assurance framework for ensuring that data submitted to regulators can be reconstructed and reanalyzed if needed (^[9]). It does not itself guarantee that the science is correct, but rather that whatever was done has been documented accurately.

Historically, GLP was introduced through regulations after high-profile cases of laboratory fraud. In the mid-1970s, U.S. investigations (such as the FBI and FDA joint probe into Industrial Bio-Test Laboratories) revealed that contract testing firms had falsified data in animal studies. These revelations led Congress to require FDA (and EPA) to establish rules ensuring auditable study practices (^[10]). Accordingly, 1978 saw the issuance of FDA’s GLP final rule (43 FR 60013, Dec. 22, 1978) (^[3]). Similar rules by EPA (under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and Toxic Substances Control Act (TSCA)) took effect around that time, allowing EPA to enforce GLP for chemical testing. Today, GLP is one of the most harmonized regulatory areas globally: In 1992 the OECD adopted mutually-accepted GLP standards, meaning data from OECD-GLP labs in member countries are recognized worldwide (^[4]) (^[11]).

GLP complements other “good practices” in regulated industries. For example, Good Manufacturing Practice (cGMP) (21 CFR Parts 210-211) governs pharmaceutical production, Good Clinical Practice (GCP) (21 CFR Part 312, ICH E6) governs clinical trials, and Good Tissues Practice (cGTP) covers human cell products. GLP is distinct in applying to preclinical nonhuman studies and tackles lab processes and documentation rather than product quality per se. Notably, GLP studies often serve as evidence in INDs and NDAs much as GMP records support drug quality. The academic and regulatory communities sometimes debate the weight given to GLP compared to peer-reviewed literature (discussed later), but GLP remains the legally mandated standard for data used in FDA submissions (^[7]) (^[5]).

The scope of 21 CFR 58 is clearly laid out in §58.1. It explicitly includes most nonclinical safety studies involving test and control articles under FDA’s jurisdiction (^[3]). Accordingly, Part 58 applies only where a regulated test article is studied prospectively in a nonclinical laboratory safety study under the definitions and exclusions in § 58.3, and the study supports, or is intended to support, an FDA research or marketing permit application. Importantly, §58.3 defines “Nonclinical laboratory study” to exclude human trials, clinical studies, field trials in animals, and basic exploratory research (^[12]). FDA’s 1981 Q&A clarifies borderline cases: for instance, it explicitly states that purely analytical method validation trials do not require GLP, whereas target-animal safety studies (e.g. overdose or tissue residue studies for veterinary drugs) do (^[13]).

In short, Part 58 applies when a prospective nonclinical laboratory safety study involving an FDA-regulated test article supports, or is intended to support, an FDA research or marketing permit application and falls within the rule’s definitions and exclusions (^[14]). Examples include acute and chronic toxicity studies, carcinogenicity bioassays, reproduction/development studies, genotoxicity assays, pharmacokinetics (toxicokinetics) studies, medical device biocompatibility tests (e.g., sensitization, implantation, pyrogenicity), and more. Conversely, basic R&D studies to screen compounds or determine chemistry/structure generally do not fall under GLP. Clinical research on humans is governed by GCP, not GLP.

Table 1. Subparts of 21 CFR Part 58 and Key Requirements

(Table 1 adapted from 21 CFR 58 text (^[3]) (^[17]).)

Scope (§58.1): 21 CFR 58.1 states that GLP applies to studies supporting FDA research or marketing applications, including drug, device, additive, and biologic approvals (^[3]). By listing statutory sections, the rule clarifies GLP data underpin requirements like safety (e.g. for NDA §512, medical device 510(k) or PMA, food additives, etc.) (^[3]). It explicitly requires compliance for safety data filed under FD&C Act and PHS Act provisions. Importantly, the rule does not cover submission of clinical efficacy or performance data. The 1987 Senate/FDA amendment note (#52 FR 33779) confirms this scope. The 1981 FDA Q&A confirms that analytical validation alone (e.g. purity testing methods) is not under GLP (^[13]), but that studies like animal safety in target species for veterinary or pesticides are included (^[18]). In practice, sponsors should assess whether each prospective nonclinical safety study falls within Part 58’s scope; relevant noncompliant study information may still need to be provided with an explanation where applicable (^[1]; 21 CFR § 812.27).

Definitions (§58.3): Key terms are defined for clarity. In particular:

• Test Article: broadly any regulated item (drug, additive, device, etc.) being tested (^[19]).
• Control Article: anything given to test systems for comparison (e.g. placebo, carrier) (^[20]).
• Nonclinical Laboratory Study: any in vivo or in vitro study in non-human test systems under controlled conditions, aimed at determining safety (^[12]). This definition explicitly excludes human clinical or field trials and excludes “basic exploratory studies” not intended for submissions (^[12]).
• Protocol: The written plan for a particular study. (Other terms like “archiving” and “test system” are also defined but follow ordinary usage.)

The definition makes it clear that GLP focuses on production of safety data. For example, pharmacokinetic studies for new drugs are GLP if intended to support an IND, but not if done as purely research. Similarly, acute toxicity tests for a pesticide registration are GLP, but environmental “field” monitoring is not.

Applicability to Grants (§58.10): GLP is stipulated for all nonclinical studies supporting FDA applications, even if conducted under federal grants or contracts (unless specifically waived) . The rule does not allow avoidance of GLP by calling a study a grant project.

Inspection (§58.15): FDA (specifically the Office of Regulatory Affairs/Bioresearch Monitoring (BIMO) program) may inspect any GLP lab any time to verify compliance (^[21]). FDA’s public resources note that inspections are routine at facilities performing submitted safety studies (^[21]). FDA publishes lists of GLP laboratories (e.g. archives from 1989–2000 and active lists 2000–2025 (^[22])) to document labs that have been inspected or have GLP registration. In short, GLP is actively enforced by FDA – it is not a voluntary guideline but a compliance requirement.

GLP requires every testing facility to have a clear organizational structure and personnel to oversee studies:

• Facility Management (§58.31): Each lab (testing facility) must have management responsible for implementing GLP. Management must ensure compliance, allocate resources, and designate roles (^[23]).
• Testing Facility Management (§58.29, 58.31): The regulations expect at least one senior person to ensure GLP implementation. In large companies, multiple units may be involved (e.g. sample management, animal care).
• Study Director (§58.33): Every nonclinical study must have a single Study Director responsible for the overall conduct of that study (^[23]). The Study Director designs the protocol, obtains approvals, directs conduct, and finalizes the report. Their signature on the report attests that data are correct and complete. The Study Director cannot delegate core responsibilities (e.g. data review, final report signoff).
• Personnel Qualifications (§58.29, 58.33): Individuals must have education or training commensurate with their duties. Managers and study directors usually have science backgrounds. Support staff (technicians, pathologists) must be qualified to perform their tasks.
• Quality Assurance Unit (QAU) (§58.35): A very important provision is that the facility must have an independent Quality Assurance Unit (QAU) that is separate from study conduct. The QAU periodically inspects and audits GLP compliance for protocols, raw data, and final reports (^[8]). For instance, the QAU checks that all SOPs are followed, equipment is calibrated, and that the Final Report faithfully presents the data. The 1981 FDA Q&A notes that the Quality Assurance staff need not share scientific expertise for all specialized tasks (e.g. pathology) but must audit processes and records for compliance (^[24]). The QAU reports integrity-threatening problems to management and the study director immediately, submits periodic written status reports, reviews the final report, and signs the required statement included with that report. Part 58 does not prescribe an annual inspection cadence or expressly give the QAU stop-work authority (^[25]).

Together, Subpart B ensures personnel accountability. For applicable nonclinical studies, Part 58 establishes organizational and quality-assurance requirements. A prior FDA inspection is not a universal prerequisite to submitting relevant nonclinical information; for example, IDE submissions must disclose whether nonclinical studies complied with applicable GLP requirements and explain noncompliance, without omitting relevant studies for that reason (^[26]). The multiple layers—management, study director, and QAU—create checks that help detect and address errors or misconduct.

The physical environment of the testing facility is regulated to prevent external variables from affecting study integrity:

• General Requirements (§58.41): Facilities must be of suitable size and construction to accommodate separate areas for animal care, test operations, storage, and archiving. Preventing cross-contamination is emphasized (e.g. separate areas for animal rooms vs. necropsy vs. data recording). Workflows must allow specimens and data to flow in a traceable manner.

• Animal Care Facilities (§58.43): The provision requires sufficient animal rooms or areas, as needed, for separation of species or test systems, isolation of projects, quarantine, and housing. It also requires separate areas for biohazardous studies, appropriate disease-isolation areas, and waste collection, disposal, or sanitary storage designed to minimize vermin infestation, odors, disease hazards, and environmental contamination (^[27]).

• Animal Supply Facilities (§58.45): This section requires storage areas, as needed, for feed, bedding, supplies, and equipment. Feed and bedding storage must be separate from animal housing and protected against infestation or contamination. Separately, §58.90(b) requires newly received animals from outside sources to be isolated and have their health status evaluated in accordance with acceptable veterinary medical practice (^[14]).

• Test Article Handling Areas (§58.47): Separate, secure areas must exist for receiving, identifying, storing, and preparing test/control substances. Labeling procedures are required so that each article’s identity, batch, and origin are known. The intent is to avoid mix-ups that could invalidate a study.

• Laboratory Operation Areas (§58.49): Separate laboratory space must be provided, as needed, for the routine and specialized procedures required by nonclinical laboratory studies (^[28]).

• Specimen/Data Storage (§58.51): There must be conditions and space for storing study specimens (e.g. tissues) and data records. Specimens (slides, tissue blocks, etc.) should be kept under conditions that preserve their integrity. Data archives should be safe from fire or unauthorized access.

These facility rules ensure that the test systems (animals, tissues, in vitro systems) are maintained properly and that test articles are handled correctly. For instance, SOPs for feeding and housing animals (§58.90) are designed to prevent malnutrition or stress from skewing toxicity results (^[15]). These controls are intended to help prevent animal-care conditions from affecting study integrity.

Equipment used in studies (e.g. analytic instruments, pipettes, autoclaves) must be suitable and maintained:

• Design (§58.61): Equipment should be of a type at least as sophisticated/accurate as the study demands. For example, if a study measures drug concentration in serum, the laboratory must have analytic equipment capable of the necessary detection limits.
• Maintenance and Calibration (§58.63): Equipment must be regularly calibrated and maintained per written procedures. All calibrations and malfunctions are documented. For instance, a chromatography machine must be calibrated with known standards at defined intervals. If a device breaks, the facility must determine if affected data are still valid.

These provisions ensure instrument accuracy and thus reliable data. Numerous FDA inspection reports cite failures where “un-calibrated” scales or thermometers called into question study validity. Part D ties into Part C by requiring equipment itself to be placed in appropriate environments.

This subpart governs day-to-day laboratory operations:

• Standard Operating Procedures (§58.81): GLP requires written SOPs for all routine processes that could affect study results or integrity. Examples include animal care, cage cleaning, chemical waste disposal, sample collection, data recording, chromatograph operation, analytical methods, etc. Management must be satisfied that SOPs are adequate to ensure data quality and integrity; significant SOP changes require written management authorization. Part 58 does not require QAU approval of every SOP. SOPs ensure consistency; for instance, “Procedure for Hematology Sample Collection” prevents variation between technicians.

• Reagents and Solutions (§58.83): Labeling and documenting the preparation of chemicals used in tests is mandatory. Reagents must be identified by name and strength, and SOPs should describe preparation methods. The 1981 Q&A clarifies that even things like lab solutions used in assays fall under this requirement (^[29]).

• Animal Care (§58.90): GLP reminds that animal husbandry itself can influence results. Subsection 58.90(a) states there shall be SOPs for housing, feeding, handling, and care of animals (^[15]). Frequent checkpoints of environment (cages, food quality), health records, and humane euthanasia methods are part of GLP animal care. Incoming animals from external sources “shall be isolated” to ensure they are disease-free before mixing with the colony (^[15]). This prevents, for example, an infectious outbreak from contaminating an ongoing carcinogenicity study over years.

• Test System Observations: During a study, test systems must be monitored in conformity with the protocol, and data must be recorded promptly, attributable to the person making or entering the record, and corrected without obscuring the original entry. Part 58 does not impose a universal requirement to perform a necropsy or require the study director to attribute the cause of every unexpected animal death; those actions may instead be required by the protocol, scientific methods, or applicable animal-care requirements (^[30]).

Together, Subpart E SOPs ensure consistency and traceability in day-to-day operations. Importantly, SOPs must be followed exactly, and any deviation (even justified protocol changes) must be documented and explained. FDA inspectors regularly check SOP records and may cite deficiencies if, for instance, a lab has an ‘uncontrolled’ method for preparing dosing solutions.

GLP mandates strict handling of the materials being tested:

• Characterization (§58.105): Prior to a study, the identity and relevant properties of each test or control substance must be determined and documented. For a new chemical, this may include purity, stability, lot analysis, chemical identifiers (CAS number), and proposed dosage form. Even an FDA-approved drug used as a test article must have its label, strength and expiration documented. The idea is that known impurities or characteristics are understood so they don’t confound results.

• Handling (§58.107): Procedures must exist for how test articles are received (documenting date, quantity, storage conditions), labeled, stored, and transferred or mixed for use. For instance, if dosing animals from a solution, the mixing protocol (solvent, concentration, testing interval) must be fixed.

• Mixtures (§58.113): Any compound created by mixing a test article with a carrier (e.g. suspending a powder in corn oil) must be documented for content and stability.

These rules are intended to prevent mix-ups or degradation of test substances. Regulatory reviewers can use the documented test-article properties to assess whether they may have affected study results (^[31]).

Protocol (§58.120): Before starting, each study must have an approved protocol document specifying objectives, design, methods, and materials. The protocol includes details like number of animals per group, dosing regimen (dose levels, route, timing), observations schedule, measurements, and endpoints. It is often accompanied by attachments such as clinical pathology data forms or necropsy schemas. Under §58.120, it must identify the sponsor and the date of sponsor approval, and it must be dated and signed by the study director. Any amendments must be dated, signed by the study director, and maintained with the protocol; a QAU review may be a facility practice but is not the protocol-approval requirement stated in §58.120 (^[32]).

The Conduct (§58.130): Once underway, the study must follow the protocol. Tasks like dosing animals, collecting samples, and running analyses occur by the described schedule. All operations are recorded as they happen (contemporaneous data logging). GLP requires that all changes (even minor) to protocol be recorded with justification (e.g. if an animal is lost, what was changed). The Study Director ensures compliance day-to-day. If new information arises (e.g. dosing error), the Director decides on halting or modifying the study.

Thus, Subpart G enforces that a study is “planned in advance and executed as planned.” This is crucial so that the Final Report accurately reflects what was done. Many GLP issues arise when labs conduct multiple overlapping studies without clear demarcation, or when data are backfilled; GLP explicitly forbids such practices. Auditors check that data entry is contemporaneous and that raw data (e.g. lab notebooks, instrument printouts) match the protocol.

A cornerstone of GLP is documentation:

• Final Report (§58.185): For each completed study, a comprehensive final report must be prepared, integrating all data and describing the conduct and results. The regulation enumerates sections that should be in the report: identification of the facility and dates; objectives/procedures (with mention of any protocol changes); statistical methods; identification of test/control articles (with specifications); method descriptions; test system description (e.g. animal numbers, sex, source, weights); dosage regimen; observations and results; discussion and conclusions (^[33]). Essentially, a GLP final report is much more detailed than a typical journal article; it is written so that an independent reviewer can understand exactly how the study was done and verify the findings.

• Data Storage and Retrieval (§58.190): All records (raw data sheets, charts, specimens) must be archived in an environment protecting from damage (firesafe, climate-controlled, etc.). Records must be indexed for retrieval by study number or other identifier. If an inspection or question arises, a sponsor must be able to produce all raw data from archives.

• Retention of Records (§58.195): Records retention is legally mandated. Except for specified fragile materials, documentation records, raw data, and specimens must be retained for whichever applicable period is shortest: at least 2 years after FDA approves the application supported by the study (with IND and IDE records governed by the five-year provision), at least 5 years after the study results are submitted to FDA, or, if the study is not submitted, at least 2 years after the study is completed, terminated, or discontinued (^[16]). The rule does not supersede longer retention obligations imposed elsewhere. This retention framework enables FDA to inspect the underlying study records when needed.

Failure to maintain complete archives (or any loss of original records) is a major violation; it could jeopardize an entire application’s credibility.

Subpart K provides enforcement teeth. If GLP compliance is knowingly ignored, FDA can disqualify the facility:

• Purpose (§58.200): The rule explicitly allows FDA to exclude from consideration any studies from a facility that failed GLP, until it is proven that noncompliance did not affect data integrity (^[34]). In practice, this means any studies conducted by a disqualified lab after the violation date cannot be used in any FDA submission. Disqualification also acts as an incentive for sponsors: if a CRO fails GLP, a sponsor must find another lab or repeat studies.

• Grounds (§58.202): “Objectionable conditions” that undermine GLP (e.g. falsified records, severe deviations, intentional data gaps) can trigger disqualification. The warning letter or enforcement notice will detail the violations.

• Process: The facility and any sponsor must be notified and given a chance to remedy. If contested, FDA holds a hearing, then issues a final order. After correction, a facility may petition for reinstatement.

This mechanism is rare but important. It underscores that GLP compliance is not optional: data from a GLP failure lab can be legally ignored, meaning huge financial losses if a drug application is denied due to inadequate safety data. It also allows FDA to demand re-analysis or new studies.

Table 2 below provides a comparative glimpse of GLP regulations across jurisdictions (FDA, EPA, OECD).

Table 2. Comparison of GLP Regulatory Frameworks

(Table 2 sources: 21 CFR Part 58, OECD GLP pages (^[5]), OECD MAD information (^[4]), EPA inspector manual (^[35]), ILAR Journal)

The table highlights that FDA GLP and international GLP frameworks overlap in important ways. Under OECD MAD, qualifying nonclinical safety studies on chemicals may be accepted across participating jurisdictions when the applicable OECD Test Guideline, OECD GLP, facility-inspection, national-monitoring-programme, and product-scope conditions are met. Whether a Canadian study supports a particular FDA submission depends on FDA's applicable requirements and review of that submission (^[36]).

A central purpose of GLP is to safeguard data integrity. Part 58 explicitly requires that raw data be recorded at the time of observation (“contemporaneously”) and be attributable. This means each data page or printout should be initialed and dated by the person making the entry. If corrections are needed, they must be made by a single line-through (no white-out), with justification. These practices help prevent and reveal falsification.

The independent Quality Assurance Unit (QAU) is a principal check on study integrity. The QAU must inspect each nonclinical laboratory study at intervals adequate to assure integrity and maintain signed records of those inspections. It must promptly bring integrity-threatening problems to the study director and management, review the final study report, and prepare and sign the QAU statement included with that report (^[25]). FDA investigators—not the QAU—issue Form FDA 483 inspection observations to firm management at the conclusion of an FDA inspection (^[37]). This oversight structure—management, QA, and study director—adds multiple levels at which errors or misconduct can be identified and addressed.

Additionally, GLP emphasizes trained personnel. §58.29 requires adequate staffing and qualifications so that technicians and scientists know how to perform tasks reliably. In practice, CROs dedicated to GLP invest in training programs, whereas an academic lab doing GLP for the first time must ramp up. The ILAR article notes that many academic researchers (who are more accustomed to flexible methods) must adapt to GLP rigor when a project shifts to product development (^[38]).

From the literature standpoint, the impact of GLP on data reliability has been debated. Some scholars (e.g. Myers et al.) argue that relying solely on GLP status as a quality filter is misguided (^[7]). They point out that GLP does not guarantee sound study design, sensitivity, or modern methodology; it only ensures documentation. Conversely, others (e.g. Borgert et al.) contend that GLP provides crucial basics that peer review does not – consistency and harmonization across labs (^[6]). The consensus view is that GLP adds process-level quality (audit trails, archiving) that augments but cannot substitute for scientific rigor. Part 58 mandates GLP for covered prospective nonclinical laboratory safety studies; FDA may also evaluate other evidence under the requirements applicable to the particular submission (^[1]; 21 CFR § 58.3).

Modern concerns include data integrity in the digital age. Part 58 requires retention and accessibility of required records but does not itself prescribe a universal electronic-system validation or audit-trail requirement. Part 11 applies to electronic records that are created, modified, maintained, archived, retrieved, or transmitted under FDA record requirements, as well as specified electronic records submitted to FDA. The applicable Part 58 requirements remain the predicate rules; FDA’s Part 11 guidance explains that FDA exercises enforcement discretion for certain Part 11 provisions while those underlying record obligations continue to apply (^[39]; FDA Part 11 Scope and Application guidance). The SG Systems glossary emphasizes that GLP now "intersects with electronic controls" under Part 11 (^[40]). FDA has issued guidance (and enforcement actions) on data integrity, stressing that GLP applies equally to e-records. The FDA’s draft Q&A on “Translation of GLP Study Reports” (November 2023, still in draft as of April 2026) highlights ensuring accuracy of translated electronic data (^[41]). The OECD has been particularly active in this space, publishing Advisory Document No. 22 on GLP Data Integrity (2021), No. 17 Supplement 1 on GLP and Cloud Computing (2023), and Position Paper No. 25 on GLP and IT Security (December 2024). FDA warning letters to Jiangsu Kerbio, CCIC Huatongwei, and Vedic Lifesciences document specific GLP deficiencies, including failures involving study records, data accuracy, and final reports. Those individual letters do not establish a systemwide ranking of deficiency categories (^[42]; CCIC Huatongwei warning letter; Vedic Lifesciences warning letter).

FDA’s Office of Regulatory Affairs administers GLP compliance through Bioresearch Monitoring (BIMO) inspections. These can be routine surveillance (scheduled at some labs per year) or for-cause (triggered by suspicions or violations). When FDA inspects, investigators tour the facility, interview staff, and review records. Inspection forms (e.g. Form FDA 483 listing deficiencies) are issued if violations are found. A final Warning Letter may follow if issues are serious.

Case studies illustrate the enforcement process:

• Example – Jiangsu Kerbio (China, 2025): FDA inspected Jiangsu Kerbio from January 15 through January 24, 2025, to assess its participation in GLP nonclinical medical-device studies. FDA’s July 11, 2025 warning letter describes serious Part 58 violations, including study-director failures involving protocol approval and the accurate recording and verification of experimental data. The letter is the authoritative source for the inspection’s observations and the firm’s response (^[42]).

• Interagency Coordination: An EPA “GLP Inspector’s Manual” (1985) recounts that early GLP efforts found ~40% of submitted animal study data to be inadequate, prompting EPA-FDA cooperation (^[43]). That manual also documents a 1978 interagency cooperation agreement: “CFR 160 (FIFRA) and 40 CFR 792 (TSCA) and 21 CFR 58 (FDA) ... provides for FDA and EPA cooperation in GLP monitoring (^[35]).” This shows that even 40+ years ago, FDA and EPA shared information and resources. In practice today, such cooperation continues (labs that do both drug and chemical testing are inspected jointly).

• Example – CCIC Huatongwei International Inspection (China, 2025): Inspected in January 2025, this Suzhou-based testing facility received a warning letter on June 25, 2025 for study director failures to accurately record experimental data, including animal weights, preparation, and test article administration. FDA described the data as unreliable, highlighting continued enforcement focus on Chinese GLP CROs (^[44]).

• Example – Vedic Lifesciences (India, 2026): Inspected in January 2025, this Indian laboratory received a warning letter on March 9, 2026 for failing to provide final reports with accurate and complete information for three GLP studies, raising concerns about data validity and integrity (^[45]). This case demonstrates that FDA's GLP enforcement extends globally and to diverse product categories.

• Disqualification in Practice: While official disqualifications are rarer and often not publicized outside government, FDA does maintain a registry of disqualified firms (21 CFR 58 Subpart K) through the Bioresearch Monitoring (BIMO) program. FDA publishes inspection and compliance information, but published BIMO summaries may aggregate programs and do not support a general GLP-only rate for Form FDA 483 observations or major findings. Inspection observations should therefore be evaluated in the context of the specific inspection and applicable FDA dataset rather than through unsourced percentage estimates.

Regulatory Perspective: Part 58 establishes recordkeeping, study-director, quality-assurance, inspection, and disqualification provisions intended to support the quality and integrity of safety data submitted to FDA. The regulation does not support a general numerical claim about the number of laboratories found in violation; inspection and enforcement information should be tied to identifiable FDA records or datasets (^[14]).

Industry Perspective: For sponsors and contract research organizations conducting covered nonclinical laboratory safety studies, Part 58 establishes requirements for organizational responsibilities, quality assurance, study conduct, and records. FDA evaluates submitted evidence and any disclosed noncompliance under the requirements applicable to the particular submission (^[1]).

Academic Perspective: As noted earlier, academia often views GLP differently. Researchers primarily focused on discovery may view GLP as onerous, and indeed many believe it should apply only when data support product approvals (^[38]). An ILAR Journal commentary describes GLP as initially foreign to basic research labs and only incrementally embraced when universities enter product development (^[38]) (^[46]). Some academics advocate for a “GLP-lite” approach where key elements (training, archiving) are used without full compliance, to improve reproducibility without the full GLP cost. In fact, a growing trend is academic core labs offering “GLP-compliant services” so investigators can outsource. This reflects a convergence: even academia recognizes the value in data rigor, especially amid reproducibility concerns. However, challenge remains that publishing non-GLP academic studies (in journals) often does not contribute to regulatory safety assessments, leading to questions about how regulatory agencies should weigh such data (^[47]) (^[48]).

Public Health Debate: Finally, environmental/public health advocates have sometimes criticized regulatory reliance on GLP. The BPA example (Myers et al. 2009) illustrates the argument that many independent studies showing low-dose effects were ignored because they lacked GLP status (^[47]). The counterargument (Borgert et al. 2016) is that GLP ensures at least basic quality, whereas unvetted non-GLP studies might be flawed or unreproducible (^[6]). Regulatory policy today often uses Klimisch scoring (a system that slightly favors GLP studies) to weight data quality, but EFSA and EPA guidelines clarify that GLP is necessary only for industrial studies (journals can be used if they meet Klimisch criteria) (^[6]). The bottom line is that GLP establishes mandatory requirements for covered nonclinical laboratory safety studies, while agencies may evaluate additional evidence outside Part 58 when assessing risk.

FDA publishes information and datasets concerning nonclinical laboratories inspected under GLP. Those materials can help identify facilities and enforcement actions, but they should not be used to infer global counts of GLP laboratories, inspection frequency, or GLP-only compliance rates unless the specific dataset, date range, and methodology support the calculation. Published warning letters and inspection records are more suitable for describing documented deficiencies in individual cases (^[49]).

The OECD highlights that global GLP harmonization through the Mutual Acceptance of Data (MAD) system has significant economic impact. By allowing one GLP study to be accepted in all member countries, duplication of animal testing is reduced. OECD estimates that this saves over EUR 309 million per year in test costs (^[4]). In human terms, MAD minimizes redundant animal use and speeds market access. (This figure is cited by OECD publications and implies that adherence to GLP/OECD guidelines brings societal benefit by reducing duplicative testing.)

Academic and industry publications reflect a range of expert perspectives on GLP:

• In toxicology journals, the debate centers on GLP vs. peer-review. Becker et al. (2009) and Zoeller & Vandenberg (2015) argued that GLP-based scoring (like Klimisch) can bias regulatory reviews against newer academic studies. Borgert et al. (2016) replied that GLP liaison with rigorous guidelines “promotes consistency, reliability, comparability, and harmonization” in risk assessment (^[6]). A 2014 task force document by the Society of Toxicology noted that GLP compliance does not in itself guarantee scientific validity, but is a critical component of good data quality; they recommend evaluating each study on multiple axes (GLP, peer-review level, relevance, etc.).

• Regulatory requirements establish documentation, quality-assurance, inspection, and record-retention controls for studies within Part 58's scope. GLP compliance should be considered alongside the study's design, methods, and relevance; unsupported survey results and generalized cost estimates should not be used to quantify GLP's effects (^[14]).

In summary, no substantive data contradicts that GLP documents reliability rather than assures scientific validity. The system trades perhaps some creativity or efficiency (extra paperwork) for traceability. Debate continues on how to evolve GLP (for example, integrating modern electronic tools, or alternative methods like in vitro testing) while maintaining data quality.

To illustrate GLP in action, below are two notable examples of GLP enforcement and application:

1. Historical – Industrial Bio-Test Labs (USA, 1970s). Although predating Part 58, this infamous case (IBT, 1976) remains a founding GLP lesson. The lab had falsified dozens of pesticide toxicity studies on rodents. After an FBI raid, many study records were found to be fabricated. The scandal led to federal litigation and ultimately catalyzed the entire GLP framework (^[10]). It exemplifies the why of GLP: in its absence, regulators had no way to detect or prevent outright fraud. Modern GLP (with archival of raw data and QA review) is a direct response to such abuses.

2. CCIC Huatongwei International Inspection (China, 2025). Warning Letter (June 25, 2025) (^[44]). Another Chinese testing facility cited for study director failures to accurately record experimental data, including animal weights and test article administration. This letter documents specific record-accuracy and data-integrity concerns at the inspected facility.

3. Vedic Lifesciences (India, 2026). Warning Letter (March 9, 2026) (^[45]). This Indian laboratory failed to provide accurate and complete final reports for three GLP studies. The case demonstrates that FDA's GLP enforcement extends globally across diverse markets and product categories.

Ensuring Data Integrity in the Digital Age

GLP regulations continue to evolve with technology. The digitalization of laboratory data raises questions about electronic-record controls, record retention, and whether Part 11 applies to the particular electronic records at issue. Part 11 establishes criteria for electronic records and signatures within its scope; it does not state that every electronic data-capture system in a GLP laboratory must be Part 11-compliant. Key developments and future directions include:

• OECD IT Security Guidance (2024): In December 2024, the OECD published Position Paper No. 25 on Good Laboratory Practice and IT Security, addressing cybersecurity risks in GLP computerized systems. This complements the earlier Advisory Document No. 22 on GLP Data Integrity (2021) and No. 17 Supplement 1 on GLP and Cloud Computing (2023), creating a comprehensive digital-era GLP framework (^[36]).
• Computerized Systems Validation (CSV): Computerized systems used to create or maintain required GLP records should be governed by procedures that preserve the integrity, traceability, and availability of those records. The applicable controls depend on the records and system at issue; Part 58 itself does not prescribe a universal computerized-system validation standard.
• E-Protocols and ELNs: The use of electronic lab notebooks (ELNs) continues to grow. When properly validated, ELNs can streamline GLP documentation, but agencies insist that they produce permanent, unalterable records.
• Blockchain and Audit Trails: Some suggest blockchain-like methods for immutable record-keeping. This remains speculative but aligns with GLP’s aim of data traceability.
• Data Sharing and Transparency: There is a push (through agencies like EMA and NIH initiatives) for greater transparency of preclinical data. In the future, sponsors might deposit GLP study reports in public repositories (encrypted with IP protections), which would enhance reproducibility and facilitate meta-analyses – akin to clinical trial registries.

Global Harmonization and Capacity Building

GLP remains a strong example of international cooperation. The OECD continues to update its GLP principles and FAQs, with recent guidance documents (2021–2024) addressing data integrity, cloud computing, and IT security in GLP environments. The OECD MAD system now includes all OECD members plus Argentina, Brazil, India, Malaysia, Singapore, South Africa, and Thailand as full adherents. For emerging economies, national GLP programs are being strengthened (for instance, India’s GLP National Accreditation Board now oversees many labs — though FDA’s 2026 warning letter to Vedic Lifesciences shows that enforcement gaps remain). Future implications include:

• Broader Mutual Acceptance: As chemical and biotech trade globalizes, pressure grows to have all major players fully reciprocate GLP data. The expansion of MAD adherents to countries like Malaysia and Thailand reflects this trend.
• Quality Beyond GLP: Concurrently, regulators are also emphasizing study quality beyond GLP — for example, emphasis on modern endpoints in toxicology. This may lead to an integrated framework where GLP compliance is a baseline, and study design is evaluated by additional peer review or guidance.
• Inspection Findings Must Be Read in Context: FDA warning letters describe the conditions observed at the individual facilities inspected. They may identify important compliance lessons, but these letters alone do not establish a general change in FDA enforcement intensity or a geographic enforcement trend.

GLP in New Methodologies

As toxicology develops (e.g. shifting to in vitro assays, omics, computational methods), questions arise: can GLP apply? The OECD already introduced GLP guidance for computer models (QSAR toolbox) and for in vitro tests (OECD Test Guidelines). Significant recent developments include:

• FDA’s NAMs Initiative (2025–2026): In April 2025, the FDA announced intent to replace animal-testing requirements with New Approach Methodologies (NAMs) over a 3–5 year horizon. In March 2026, FDA released draft guidance on validating NAMs as alternatives to animal testing. The FDA’s ISTAND (Innovative Science and Technology Approaches for New Drugs) program continues qualifying NAM tools for regulatory use, and the FDA’s CDER/OND has documented qualified organ-on-chip platforms including the Emulate Liver-Chip for drug-induced liver injury (DILI) assessment and CN Bio PhysioMimix (^[50]). NIH announced its intent to establish the Office of Research Innovation, Validation, and Application (ORIVA) in April 2025 and announced the office’s creation on June 15, 2026 (^[51]; NIH, June 2026).
• GLP for Non-Animal Tests: As NAMs gain regulatory acceptance, sponsors should determine the applicable evidentiary and study-conduct requirements for the specific product, application, and intended use. Part 58 applies only when a prospective nonclinical laboratory safety study involving regulated test articles falls within its defined scope; it does not establish a blanket GLP requirement for every alternative safety assay included in a submission (^[14]).
• Risk-Based GLP: Some have proposed a more risk-based GLP, where the extent of documentation could scale with study importance. This is not yet implemented but could be a future trend for efficiency, particularly as lower-risk NAMs proliferate.

Policy and Social Implications

Finally, GLP underpins societal trust in regulated products. Whenever a safety scandal erupts (e.g. a contaminated drug or device injury), investigators examine whether underlying data followed GLP. Public confidence partly rests on assurance that safety testing was rigorous. Conversely, overly rigid application of GLP has been criticized as stifling academic innovation and sometimes obscuring valid scientific findings (per the BPA controversy). Policy discussions may increasingly focus on how to modernize GLP without sacrificing its core: that data keepers can always show what they did, when, and how.

21 CFR Part 58’s Good Laboratory Practice regulations have, since 1978, served as a foundational quality standard for nonclinical safety research in the FDA-regulated arena (^[3]) (^[9]). By mandating organizational oversight, full documentation, and archiving, GLP aims to make preclinical data reliable, auditable, and globally shareable. It establishes a uniform procedural baseline intended to support the quality and integrity of covered safety data; FDA may inspect, question, or decline to consider studies when compliance or data-validity concerns arise (^[14]).

However, GLP is not a panacea. It is a framework that must be coupled with sound scientific design and modern analytical rigor. The literature highlights that while GLP ensures data traceability, it cannot guarantee scientific relevance or novelty (^[7]) (^[6]). Still, across multiple perspectives — from regulators to industry to academia — GLP is largely viewed as essential when human and environmental safety is at stake. Its harmonization under OECD facilitates international trade of test data and spares redundant animal testing (^[4]) (^[5]).

Looking forward, GLP will continue to evolve. Electronic recordkeeping, novel methods, and global partnerships will shape how Part 58 is implemented. The FDA’s draft guidance on translated study reports (originally published November 2023, still in draft status as of April 2026) (^[41]) and interagency cooperation continues the trend of adapting GLP to a connected world. The OECD’s 2024 IT security position paper and the FDA’s 2025–2026 push toward New Approach Methodologies (NAMs) represent the next frontier — ensuring that GLP principles of traceability and reproducibility extend to organ-on-chip platforms, AI-driven toxicology, and cloud-based laboratory systems. Importantly, as regulatory science grapples with new challenges (e.g. gene therapies, artificial intelligence in drug discovery, and the replacement of animal testing), the spirit of GLP — a disciplined system of verifiable data production — is likely to remain a steady cornerstone. In sum, 21 CFR Part 58 embodies the regulatory commitment to trustworthy preclinical data, ensuring that when a product reaches market, its safety claims are built on an unshakable foundation of documented science (^[5]) (^[8]).

• Electronic Code of Federal Regulations, Title 21, Part 58 – Good Laboratory Practice for Nonclinical Laboratory Studies (^[14]).
• U.S. Food and Drug Administration, Nonclinical Laboratories Inspected under Good Laboratory Practices. FDA inspections & compliance webpages (^[21]) (^[22]).
• FDA, Questions & Answers: Good Laboratory Practice Regulations (1981, updated 2007). Office of Regulatory Affairs – Nonclinical Laboratories. (Provides interpretive guidance on GLP scope and provisions) (^[13]).
• SG Systems Global, “21 CFR Part 58 – Good Laboratory Practice (GLP) (Glossary)”, updated Oct 2025 (industry overview of GLP requirements) (^[9]) (^[8]).
• OECD, Good Laboratory Practice (GLP) and Compliance Monitoring. OECD website (describes OECD GLP principles, responsibilities, MAD system) (^[5]) (^[52]).
• OECD, The Mutual Acceptance of Data (MAD) System. OECD website (explains international GLP data sharing and cost savings) (^[4]).
• P. Myers et al., “Why Public Health Agencies Cannot Depend on Good Laboratory Practices as a Criterion for Selecting Data: The Case of Bisphenol A”, Environmental Health Perspectives 117(3):309–315 (2009). (Critical analysis of GLP vs non-GLP data validity) (^[7]) (^[48]).
• C.J. Borgert et al., “Does GLP Enhance the Quality of Toxicological Evidence for Regulatory Decisions?”, Toxicological Sciences 151(2):206–217 (2016). (Discussion of GLP’s role in study quality) (^[6]).
• B. Bolon et al., “Good Laboratory Practice in the Academic Setting: Fundamental Principles for Nonclinical Safety Assessment and GLP-Compliant Pathology Support…”, ILAR Journal 59(1):18–35 (2018). (GLP principles and adoption in academia) (^[38]) (^[46]).
• U.S. Environmental Protection Agency, Good Laboratory Practice Compliance Inspections of Labs Conducting Health Effects Studies: Inspector’s Manual (1985). (Historical context of FDA/EPA GLP coordination) (^[35]).
• Federal Register, Vol. 88, No. 224 (Nov 22, 2023), Notice of Availability of Draft Guidance: “Translation of GLP Study Reports: Q&A”. (FDA guidance on GLP study report translation, still in draft as of April 2026) (^[41]).
• FDA Warning Letter – Jiangsu Kerbio Medical Technology Group Co. – July 11, 2025 (inspection of GLP lab for medical-device studies) (^[42]).
• FDA Warning Letter – CCIC Huatongwei International Inspection Co., Ltd. – June 25, 2025 (GLP violations at Chinese testing facility) (^[44]).
• FDA Warning Letter – Vedic Lifesciences Pvt. Ltd. – March 9, 2026 (GLP violations at Indian laboratory) (^[45]).
• OECD, Position Paper No. 25: Good Laboratory Practice and IT Security (December 2024). Addresses cybersecurity risks in GLP computerized systems.
• FDA/CDER/OND, “Experience with New Approach Methodologies (NAMs) Submissions”, International Journal of Toxicology (2026). Documents qualified organ-on-chip platforms for regulatory use (^[50]).
• 21 CFR Chapter I Authority footnote (43 FR 60013, Dec. 22, 1978); 52 FR 33779 (Sept. 4, 1987) (codifies Part 58) (^[3]).
• OECD GLP Series publications and national GLP monitoring program lists (see OECD website links) (^[5]) (^[52]).
• FDA GLP Lab Accreditation/Disqualification pages; EPA-FDA Interagency GLP Agreement (see BIMO web resources).
• Additional journal articles, regulatory guidance documents, and FDA compliance reports as cited above.