How Long to Keep Clinical Trial Records: Retention Periods and Archiving Obligations

At a glance

• There is no single GCP retention period. ICH E6(R3) deliberately does not state one, and ties retention to applicable regulatory requirements instead.
• Under 21 CFR Part 312 the period is two years, for both investigators and sponsors, but what the two years runs from is the part teams get wrong.
• Under EU Regulation 536/2014 the clinical trial master file is archived for at least 25 years after the end of the trial, binding on sponsor and investigator alike.
• Where obligations overlap, E6(R3) states the rule directly: whichever is longest. A multi-region trial cannot destroy on the shortest clock.
• Duration is only half the obligation. Records have to stay complete, readable, readily available and directly accessible for the whole period, which is a harder promise for an eTMF than for a box.

Why there is no single answer

The most common mistake in this area is asking “what is the GCP retention period” and expecting a number. GCP does not supply one, and that is deliberate.

ICH E6(R3) frames the duty as a pointer to whatever law applies to your trial. The investigator or institution should retain the essential records for the required retention period in accordance with applicable regulatory requirements, or until the sponsor informs the investigator or institution that these records are no longer needed, whichever is the longest (ICH E6(R3) §2.12.12).

That clause does three things at once, and each matters operationally.

It makes the period jurisdictional. The number comes from the regulation governing the trial, not from GCP.

It gives the sponsor a ratchet in one direction only. A sponsor can tell a site to keep records longer than the legal minimum, and that instruction binds. A sponsor cannot tell a site to keep them for less.

And “whichever is the longest” resolves the multi-region case. When a trial runs under more than one framework, the obligations do not average out and you do not get to pick the most convenient. The longest applicable period governs.

The United States: two years, but two years from what

For trials under a US IND, Part 312 states the period twice, once for each party, and both times as two years.

For sites, an investigator shall retain records required to be maintained under Part 312 for a period of two years following the date a marketing application is approved for the drug for the indication for which it is being investigated, or, if no application is to be filed or if the application is not approved for such indication, until two years after the investigation is discontinued and FDA is notified (21 CFR §312.62).

For sponsors, a sponsor shall retain the records and reports required by that part for two years after a marketing application is approved for the drug, or, if an application is not approved for the drug, until two years after shipment and delivery of the drug for investigational use is discontinued and FDA has been so notified (21 CFR §312.57).

The number is easy. The trigger is where teams go wrong, in three recurring ways.

It does not run from the end of the trial. The clock starts at marketing approval, which may be years after your site closed out. A trial that finishes in 2026 and supports an approval in 2031 carries records into 2033.

The fallback branch is conditional on notifying FDA. If no application is filed, or the application is not approved for that indication, the two years runs from discontinuation and notification. A team that quietly stopped an investigation without notifying has not started its clock.

Two years is a floor, not a plan. Read alongside §2.12.12, if a sponsor has instructed longer retention, or another jurisdiction’s requirement applies to the same trial, two years is not your answer.

So when the question is put in its common exam form, “the FDA requires retention of investigational drug study records for”, the accurate answer is two years, with the qualifier that it runs from marketing approval, or from discontinuation plus notification where there is no approval. The bare “two years” is only half the rule.

The European Union: at least 25 years

The contrast with the US is stark enough that assuming one framework while operating in the other is a serious error.

Under Regulation 536/2014, unless other Union law requires archiving for a longer period, the sponsor and the investigator shall archive the content of the clinical trial master file for at least 25 years after the end of the clinical trial (EU Regulation 536/2014 Article 58).

Three features distinguish it from the US rule. It runs from the end of the trial, which is knowable at the time rather than contingent on a future approval. It binds sponsor and investigator alike, in the same sentence. And it is expressed as a floor, with “unless other Union law requires archiving for a longer period” written into the text.

This is not a conflict to be reconciled. The two regimes govern different trials, and where a trial falls under both, §2.12.12 has already told you which applies: the longest.

Retention is a condition, not just a duration

Keeping records for the right number of years while letting them degrade satisfies nothing. E6(R3) states the condition alongside the duty: the sponsor and investigator or institution should retain the essential records in a way that ensures that they remain complete, readable and readily available, and are directly accessible upon request by regulatory authorities, monitors and auditors (ICH E6(R3) §C.2.6).

Unpack that into four separate tests, because a record set can pass one and fail another.

Complete. The set, not the individual document. A file missing the amendment that superseded a document is incomplete even though every document in it is intact.

Readable. Legibility over decades is a real constraint for thermal paper, for handwriting, and for file formats whose readers no longer exist.

Readily available. Retrievable on a realistic timescale. Records in deep offsite storage with a multi-week recall are not readily available in any sense an inspector would accept.

Directly accessible on request. Note who the request comes from: regulatory authorities, monitors and auditors. Access arrangements that depend on a departed employee, a lapsed vendor contract or a password nobody recorded fail here.

The EU framework states the same expectation for the trial master file, requiring it to be readily available and directly accessible upon request to the competent authorities of the Member States (EMA TMF guideline, 2018).

Archiving is not the same as storage

Archiving is a controlled state with properties storage does not have, and the distinction is where audit findings concentrate.

Archived records should be protected from unauthorised changes in order to maintain authenticity (EMA TMF guideline, 2018). Once a record is archived it should not be alterable in the ordinary course. A live system where a document can still be edited is not an archive, whatever the folder is called.

Archives should be under the control of the named individuals responsible for archiving (EMA TMF guideline, 2018). Named is the operative word. An archive that is nominally owned by a department rather than a person tends to be owned by nobody, and the gap surfaces when the one person who understood the arrangement leaves.

Electronic records and the eTMF

The obligation is media-neutral, which sounds permissive and is in fact demanding. The media used to archive the content of the clinical trial master file shall be such that the content remains complete and legible throughout the retention period (EMA TMF guideline, 2018, quoting the Regulation). The content may be retained electronically and must remain readily available (EMA TMF guideline, 2018).

Put that against a 25-year obligation and the scale of the commitment becomes clear. Twenty-five years is longer than most software companies exist and far longer than a typical vendor contract. An eTMF obligation is therefore a succession problem as much as a storage one, and it raises questions worth answering while you still have the vendor relationship:

• What happens to the archive if the vendor is acquired, changes its terms, or ceases trading?
• In what format does the content leave the system, and is that format readable without that vendor’s software?
• Does export preserve the audit trail and the metadata, or only the documents?
• Who inside your organisation is named as responsible, and what happens when they leave?

Migration between systems is itself a retention event. Content that arrives incomplete, or illegible, or stripped of its audit trail has failed the §C.2.6 condition regardless of how carefully the project was run.

Custody has to be named, and kept current

Retention obligations run for years or decades; the people who accepted them do not stay that long. E6(R3) puts a standing duty on the site to keep the sponsor told who holds the records. The investigator or institution should keep the sponsor informed of the name of the person responsible for maintaining the essential records during the retention period, for example when the investigator site closes or an investigator leaves the site (ICH E6(R3) §2.12.13).

Two details make this more than an administrative nicety. The obligation is continuing, not a single startup declaration, so the sponsor should expect updates as staff and sites change. And the examples given are the two hardest cases: a site that closes, and an investigator who leaves. Both are moments when custody is most likely to lapse silently, because the records outlast the arrangement that produced them.

In practice this is the obligation most often discovered to be broken, and it is usually discovered at the worst time, when someone needs a record from a site that shut years ago and no current name is on file.

What you are actually retaining, and why

Retention applies to the essential records, and the reason they are kept explains what condition they have to be kept in. Essential records should be available to regulatory authorities, monitors, auditors and IRBs or IECs as appropriate upon request, to enable appropriate evaluation of the trial conduct in order to ensure the reliability of trial results (ICH E6(R3), Annex 1).

Read that backwards and it defines the test. The purpose is to let someone who was not there reconstruct how the trial was run and judge whether its results can be relied on. A record set that cannot support that reconstruction has failed, however diligently it was stored.

That framing settles a question teams often litigate internally: whether something marginal is worth keeping. If its absence would leave a gap in the reconstruction of trial conduct, keep it. The standard also states the baseline plainly, that essential records should be retained securely by sponsors and investigators for the required period in accordance with applicable regulatory requirements (ICH E6(R3), Annex 1). Securely covers both loss and unauthorised alteration.

The practical rule for a multi-region trial

Identify every framework the trial falls under. Take the longest applicable period. Add any longer instruction the sponsor has issued. Then write down the reasoning, because the retention decision will outlive the people who made it, and in twenty years the only defence of a destruction date is the record of how it was calculated.

Destroying early is the one error in this area that cannot be corrected later.

Sources

• ICH E6(R3) Good Clinical Practice, version R3
• 21 CFR Part 312, Investigational New Drug Application (current as of 4/14/2026)
• Regulation (EU) No 536/2014 on clinical trials on medicinal products for human use
• EMA Guideline on the content, management and archiving of the clinical trial master file (2018)