ICH E6(R3): The New Good Clinical Practice, and Training Sized to the Task

Good Clinical Practice was rewritten for the first time in more than two decades, and the new version has been in effect since 23 July 2025. A second annex follows on 15 January 2027. For anyone responsible for site staff training in clinical research, the most consequential change is not a new requirement at all — it is permission to stop doing something.

The dates, precisely

ICH E6(R3) arrived in two parts, and conflating them causes confusion:

• Principles and Annex 1 — endorsed at ICH Step 4 on 6 January 2025, finally adopted by the CHMP on 12 December 2024, and in effect from 23 July 2025.
• Annex 2 — endorsed at Step 4 on 3 June 2026, adopted by the CHMP on 25 June 2026, and coming into effect on 15 January 2027.

Structurally the guideline opens with an Introduction and then Section II, the eleven Principles of ICH GCP. The annexes provide the basis for interpreting those principles: Section III, Annex 1 covers the IRB/IEC, the investigator, the sponsor and data governance; Section IV, Annex 2 covers the IRB/IEC, the investigator and the sponsor, with Appendices A to C and a glossary.

Training sized to the task

The provision that should change practice is Section 2.3.2. Where an investigator delegates trial-related activities, the delegates must be appropriately qualified and adequately informed about relevant aspects of the protocol — and trial-related training should correspond to what is necessary to enable them to fulfil their delegated trial activities.

Read that against standard practice. Many sites put every person listed on a delegation log through the same full protocol training package, regardless of what they actually do on the study. E6(R3) does not ask for that. It asks for training matched to the delegated activity — which means a phlebotomist drawing samples and a sub-investigator making eligibility decisions should not necessarily receive the same training, and documenting that they did is not evidence of compliance.

The same logic runs through delegation documentation. Section 2.3.3 states that documentation of delegation should be proportionate to the significance of the trial-related activities, and that delegation documentation may not be required at all where the activities form part of ordinary clinical practice.

Proportionality is a principle, not a concession

Principle 7 requires that processes be proportionate to risk and that they avoid unnecessary burden on participants and investigators. Principle 7.4 states that the sponsor should not place unnecessary burden on participants and investigators.

It is worth being precise here, because this provision is frequently over-read. E6(R3) does not contain an explicit statement that training should avoid duplication or repetition. What it says is that training should correspond to what is necessary for the delegated activity (2.3.2), that documentation should be proportionate to significance (2.3.3), and that processes generally should avoid unnecessary burden (Principle 7). Those support a right-sized approach to training; they are not a blanket licence to skip it, and anyone relying on them should be able to justify the judgement rather than cite a rule that does not exist.

Qualification remains the floor

None of the above softens the baseline. Principle 5.1 requires that individuals be qualified by education, training and experience to perform their respective tasks. Section 2.1.1 requires the investigator to be qualified by education, training and experience to assume responsibility for the proper conduct of the trial, and to provide evidence of this. Section 2.1.2 requires the investigator to know how to use the investigational product appropriately, as described in the protocol and the Investigator’s Brochure.

The obligation is mirrored on the sponsor side. Section 3.7.1 requires that each investigator the sponsor selects be qualified by education, training and experience — so investigator qualification is something the sponsor has to satisfy itself about, not something it can accept on assertion.

And Section 2.2.2 addresses resourcing directly: the investigator needs sufficient time, staff and facilities, including an adequate number of available and qualified staff. A qualified investigator at an under-staffed site is a GCP problem, not a scheduling one.

Oversight scales with the data

Delegation is expressly permitted. Section 2.3.1 allows the investigator to delegate trial-related activities to other persons or parties, but the investigator retains the ultimate responsibility and should maintain appropriate oversight. Crucially, the degree of oversight should depend on the nature of the delegated activities and be proportionate to the importance of the data being collected.

That is a risk-based oversight standard rather than a uniform one, and it sits alongside Principle 10.1, which confirms that delegating parties retain overall responsibility for their respective activities. Two further specifics: under Section 2.7.2(d) the investigator may delegate safety reporting activities but keeps overall responsibility for participant safety; and under Sections 2.10.2 and 2.10.3, where a pharmacist or other individual handles investigational product management, that person must be under the oversight of the investigator or institution, with the degree of oversight depending on factors such as the product’s characteristics and the route and complexity of administration.

Annex 2, and the planning window that is already open

Annex 2 is not a proposal. It has been endorsed at ICH Step 4 and adopted by the CHMP, with a fixed effective date of 15 January 2027. Its scope covers the IRB/IEC, the investigator and the sponsor, together with Appendices A to C and a glossary.

Two things follow. First, anyone writing training materials or standard operating procedures against E6(R3) now should check whether the area they are documenting falls within Annex 2’s scope, because material written to Annex 1 alone may need revisiting within the year. Second, the fact that both parts are already adopted means there is no uncertainty to wait out — the text exists and can be read, which makes a staged implementation plan straightforward to build rather than speculative.

Organisations that treated the 23 July 2025 date as the end of the E6(R3) transition will find they are halfway through it.

Why “evidence of this” is the operative phrase

Section 2.1.1 does not simply require an investigator to be qualified. It requires them to provide evidence of being qualified by education, training and experience. Section 3.7.1 puts the mirror-image obligation on the sponsor, which must satisfy itself that each investigator it selects meets that standard.

In practice this is where most sites and sponsors meet each other: the CV, the GCP training record, the medical licence, the evidence of prior trial experience. The shift in E6(R3) is not that these are newly required, but that the surrounding provisions on delegation and proportionality make the content of the record matter more than its existence. A delegation log naming twelve people with identical training records is now a weaker document than one showing differentiated training matched to differentiated roles — because the latter is what Section 2.3.2 describes and the former is what it quietly argues against.

The same architecture as every other regime this year

What makes E6(R3) worth reading outside clinical research is that it lands on the same design as several unrelated regimes. Training and oversight calibrated to risk, competence that has to be evidenced rather than asserted, and accountability attached to a named individual who may delegate the work but not the responsibility.

EU GMP Chapter 2 requires continuing training with its effectiveness periodically assessed, as our guide to GMP training requirements sets out. The FDA’s Quality Management System Regulation, in force since February 2026, adopted ISO 13485 clause 6.2 and its requirement that effectiveness checks be proportionate to the risk of the work — covered in our guide to the QMSR and ISO 13485 competence requirements. Pharmacovigilance attaches the duty to a named, qualified individual who may delegate tasks only under documented supervision, as set out in our guide to the QPPV role. Medical device manufacturers appoint a Person Responsible for Regulatory Compliance on the same pattern, and data integrity expectations require reviewers who genuinely understand what they are approving — see our guide to data integrity training and MHRA GxP expectations.

What to change at site level

Three practical consequences follow from the 2025 text. Map training to delegated activities rather than issuing one package to everyone on the delegation log, and record the reasoning so the judgement is auditable. Scale oversight to the importance of the data a delegate is generating, rather than applying a single monitoring intensity across all activities. And review whether delegation documentation is being produced for activities that form part of ordinary clinical practice, where Section 2.3.3 indicates it may not be required.

The one thing not to do is treat proportionality as a reason to document less without deciding more. E6(R3) replaces uniform process with justified judgement — which is more demanding of the people designing the training, not less.