Key points
- Article 58 of Regulation (EU) No 536/2014 sets a floor of 25 years for the trial master file, and the UK matched that figure on April 28, 2026.
- Twenty-five years is neither a maximum nor a global norm: the US still works from two years, Canada from 15, and ATMP traceability from 30.
- ICH E6(R3) deleted the old fixed default and widened retention from documents to essential records, metadata included.
- Archived databases must remain restorable and dynamic; static exports of dynamic data are not accepted.
- Essential documents produced the largest subcategory of major and critical findings in EMA's 2024 GCP inspections.
Retention used to be the quiet part of a trial: box the binders, sign the log, move on. Four regulatory changes between January 2025 and June 2026 ended that. The European horizon now runs to a quarter of a century, retention covers audit trails and metadata, and the confidentiality rules governing archives are being rewritten while those archives are already filling up.
💡 Need to map your retention obligations across sites? Request a personalized demo →
How long do trial records really have to be kept?
There is no single answer, and the search for one is the most common planning error. Article 58 of the Clinical Trials Regulation requires the trial master file to be archived for at least 25 years after the end of the trial, with subject medical files left to national law. The UK reached the same figure on April 28, 2026, though the transitional split matters: applications submitted before that date keep the old five-year TMF period, while participant medical files sit at 25 years for every trial.
The United States did not follow. 21 CFR 312.62(c) ties investigator retention to two years after marketing approval, or two years after the investigation is discontinued and the FDA is notified, a clock that may start decades later, or never. HIPAA documentation runs six years, Common Rule records three. Health Canada moved the other way, cutting 25 years to 15 in February 2022.
| Jurisdiction or record type | Minimum retention | Trigger point |
|---|---|---|
| EU, trial master file (CTR trials) | 25 years | End of the clinical trial |
| UK, TMF, applications from Apr 28, 2026 | 25 years | Day after conclusion of the trial |
| UK, participant medical files | 25 years | Day after conclusion of the trial |
| EU, medical devices (MDR Annex XV) | 10 years (15 if implantable) | End of investigation or last device on the market |
| EU and UK, ATMP traceability | 30 years | Product expiration date |
| US, investigator records (drugs) | 2 years | Approval, or discontinuation plus FDA notice |
| Canada, drug trial records | 15 years | Under the Food and Drug Regulations |
| France, legacy biomedical research | 15 years (40 if blood-derived) | End or early termination |
ℹ️ Note: The French figures come from the ministerial orders of November 8, 2006, and August 11, 2008. The 2026 versions of the CNIL reference methodologies MR-001 and MR-003 set when data leaves the active database and defer archiving to sectoral rules. For research that falls outside these methodologies, the CNIL assesses the period set by the data controller when it reviews the authorization request.
Every figure is a floor. The governing period is the longest applicable rule across every country where the trial ran and every market where the product may be authorized.
[[cta]]
What changed when ICH E6(R3) replaced documents with records?
E6(R3) reached Step 4 on January 6, 2025, and took effect in the EU on July 23, 2025. Two changes reshape archiving.
The first is definitional. Retention no longer attaches to essential documents but to essential records: documents and data, plus relevant metadata, in any format. Audit trails, event logs, and system metadata now sit inside the retention perimeter rather than beside it.
The second is a deletion. E6(R2) carried a default of two years after the last marketing approval in an ICH region; E6(R3) removes it. Section 2.12.12 points instead to applicable regulatory requirements, or to the moment the sponsor confirms in writing that the records are no longer needed, whichever runs longest. The guideline no longer supplies a number. Section 3.6.3(c) moves the obligation into sponsor agreements, and section 2.12.13 requires sites to name who holds the records once a site closes or an investigator leaves.
What does a compliant electronic archive look like in practice?
EMA's guideline on computerized systems is the operative text, and it is demanding: retention periods known for metadata as well as data, an inventory of records maintained, files archived read-only, and transfers run through a validated process.
Database decommissioning is where most plans fail. A dated, certified copy must be archived and restorable, including dynamic functionality, audit trails, event logs, queries, and user logs. Static formats of dynamic data are explicitly inadequate, and loss of integrity before the end of the period counts as data loss.
Investigator independence is protected throughout. Site data must remain available to the investigator after the trial, the sponsor must never hold exclusive control, and before read-only access is withdrawn at database lock, a copy including the audit trail must reach the investigator directly. An electronic data capture platform that exports complete, restorable datasets rather than flat snapshots determines how expensive that obligation becomes ten years later. The MHRA adds a named individual accountable for archiving, covering electronic records as well as paper.
💡 Comparing platforms on their export and audit-trail capabilities? Read the comparison
Who can still see the data once the trial closes?
Confidentiality is the unsettled half. The GDPR applies across the retention period, but a participant has no right to erasure of trial data where deletion would seriously impair the research, for instance by biasing safety analyses (GDPR Article 17(3)(d)), and participants should be told so.
Two developments are live. On September 4, 2025, the Court of Justice held in EDPS v. SRB that pseudonymized data is not automatically personal data for every recipient, directly relevant whenever datasets reach statisticians, central labs, or adjudication committees. EDPB guidance on pseudonymization and anonymization remains in draft, so re-identification risk should be documented rather than assumed. The CNIL's revised MR-001 and MR-003 require inclusion numbers that carry no identifying meaning, strict custody of the linkage table, and segregation of administrative and research data. The security annex applies to research initiated on or after May 23, 2026; studies already under way must document an action plan instead. Multifactor authentication becomes mandatory on January 1, 2027, for web-accessible systems. Research that cannot claim conformity with a reference methodology needs an authorization request and, in most cases, a data protection impact assessment. Aligning platform configuration with a documented security and privacy framework at study startup costs far less than retrofitting one mid-inspection.
Where do inspections actually go wrong?
EMA's GCP Inspectors Working Group reported 67 site inspections in 2024 and 335 major and critical deficiencies. Essential documents was the largest single subcategory, with 41 findings, ahead of source documentation and monitoring. Responsibility fell to the sponsor in 52.8% of cases.
The named examples are unglamorous and preventable: archiving requirements absent from sponsor–investigator contracts, inadequate archiving facilities, missing version control, no documented quality-control process for the TMF, pseudonymized data sent without encryption. Archiving fails at study startup, in the contract, long before anyone opens the archive.
[[cta]]
[[faq]]
Sources
- European Parliament and Council, Regulation (EU) No 536/2014 on clinical trials on medicinal products for human use, Article 58, 2014
- UK Parliament, The Medicines for Human Use (Clinical Trials) (Amendment) Regulations 2025 (SI 2025/538), regulation 22 and transitional provisions, made April 28, 2025, in force April 28, 2026
- UK Parliament, The Medicines for Human Use (Clinical Trials) Regulations 2004 (SI 2004/1031), regulation 31A, as amended
- US Food and Drug Administration, 21 CFR 312.62, Investigator recordkeeping and record retention, paragraph (c)
- US Department of Health and Human Services, HIPAA Security and Privacy Rules, 45 CFR 164.316(b)(2) and 45 CFR 164.530(j)
- US Department of Health and Human Services, Common Rule, 45 CFR 46.115(b)
- Government of Canada, Food and Drug Regulations, section C.05.012(4), as amended by SOR/2022-18
- European Parliament and Council, Regulation (EU) 2017/745 on medical devices, Annex XV, Chapter III, 2017
- European Parliament and Council, Regulation (EC) No 1394/2007 on advanced therapy medicinal products, Article 15(4), 2007
- French Ministry of Health, arrêté du 8 novembre 2006 (biomedical research on medicinal products) and arrêté du 11 août 2008 (other biomedical research), retention periods for sponsors and investigators
- ICH, Guideline for Good Clinical Practice E6(R3), final version of January 6, 2025, sections 2.12.12, 2.12.13 and 3.6.3
- ICH, Integrated Addendum to ICH E6(R1): Guideline for Good Clinical Practice E6(R2), sections 4.9.5 and 5.5.11, November 9, 2016
- European Medicines Agency, ICH E6 Good clinical practice, scientific guideline, effective date of E6(R3) in the EU
- European Medicines Agency, Guideline on computerised systems and electronic data in clinical trials, EMA/INS/GCP/112288/2023, March 2023
- European Parliament and Council, Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 17(3)(d) and 89, 2016
- Court of Justice of the European Union, judgment of September 4, 2025, EDPS v. SRB, Case C-413/23 P
- European Data Protection Board, Guidelines 01/2025 on Pseudonymisation, version for public consultation, January 2025
- CNIL, délibération n° 2026-050 du 19 mars 2026 (MR-001) and délibération n° 2026-051 du 19 mars 2026 (MR-003), Journal officiel of May 23, 2026
- CNIL, Recherche en santé : la CNIL met à jour et élargit le champ des méthodologies de référence 001 et 003, May 26, 2026
- European Medicines Agency, Annual Report of the Good Clinical Practice (GCP) Inspectors' Working Group (IWG) 2024, EMA/INS/GCP/37608/2025, December 15, 2025
How long does the FDA require clinical trial records to be kept?
21 CFR 312.62(c) ties investigator retention to two years after the FDA approves the marketing application, or two years after the investigation is discontinued and the FDA is notified. There is no fixed calendar period: the clock can start decades after the trial ends, or never start at all. HIPAA documentation runs six years, Common Rule records three.
Can archived clinical data be kept as PDF or CSV exports?
Not for data that was dynamic in the live system. A dated, certified copy has to stay restorable with its dynamic functionality, audit trails, event logs, queries and user logs intact. A flat export of a dynamic dataset does not meet the expectation, and losing integrity before the end of the retention period counts as data loss.
Who is responsible for archiving, the sponsor or the investigator?
Both, and the split has to be written down before anyone needs it. ICH E6(R3) moves the obligation into sponsor agreements under section 3.6.3(c), and section 2.12.13 requires sites to name who holds the records once a site closes or an investigator leaves. The MHRA expects a named individual accountable for archiving, covering electronic records as well as paper. EMA inspectors listed archiving requirements missing from sponsor-investigator contracts among their 2024 findings.
Can a participant ask for their trial data to be deleted?
In most cases, no. The GDPR applies across the retention period, but its Article 17(3)(d) sets the right to erasure aside when deletion would make the research impossible or seriously impair it. Deleting a participant's data would bias the safety analyses, so the right to erasure seldom reaches trial data. Tell participants this at consent rather than letting them discover it when they ask.
What happens to trial records when a site closes or an investigator leaves?
Someone has to be named in writing before it happens, under ICH E6(R3) section 2.12.13. Site data stays available to the investigator after the trial, the sponsor never holds exclusive control, and before read-only access is withdrawn at database lock, a copy including the audit trail reaches the investigator rather than a third party.
With over 10 years of experience working in CROs, Khalil brings deep expertise in clinical trials and a clear understanding of the daily challenges faced by research professionals. His insights are grounded in real-world operations, making his perspective both practical and strategic.
Blog & News Datacapt
News, Articles, Resources et Tutorials.
