---
title: Pre-Clinical Testing Documentation: What to Include for Bench and Animal Studies
description: Pre-clinical test evidence supports the benefit-risk analysis under MDR. Here is what to document for bench tests and animal studies in your technical file.
authors: Tibor Zechmeister, Felix Lenhard
category: Technical Documentation & Labeling
primary_keyword: pre-clinical testing documentation MDR
canonical_url: https://zechmeister-solutions.com/en/blog/pre-clinical-testing-documentation
source: zechmeister-solutions.com
license: All rights reserved. Content may be cited with attribution and a link to the canonical URL.
---

# Pre-Clinical Testing Documentation: What to Include for Bench and Animal Studies

*By Tibor Zechmeister (EU MDR Expert, Notified Body Lead Auditor) and Felix Lenhard.*

> **Pre-clinical testing documentation under MDR is the body of non-clinical evidence — bench tests, laboratory studies, simulated-use tests, and, where justified, animal studies — that supports the benefit-risk analysis of a medical device before it is placed on the market or before any clinical investigation on humans. It lives in Annex II section 6.1 of Regulation (EU) 2017/745. The section is called "pre-clinical and clinical data" and the pre-clinical part has to answer, on its own, what the device does, how it performs against its specifications, what the known hazards are, and what evidence reduces the residual risk to an acceptable level before the device ever touches a patient. MDR Annex I Chapter I requires the benefit-risk analysis and general safety obligations; MDR Article 61 ties the pre-clinical evidence into the clinical evaluation. EN ISO 14971:2019+A11:2021 frames the risk management process the evidence has to serve. A well-built pre-clinical file documents each test with a hypothesis, a method, an acceptance criterion, a result, and an explicit link back to a design input, a GSPR row, and a risk control — whether the test was run on a bench, in a tissue bath, or in an animal model.**

**By Tibor Zechmeister and Felix Lenhard. Last updated 10 April 2026.**

---

## TL;DR

- Pre-clinical evidence under MDR belongs in Annex II section 6.1 of Regulation (EU) 2017/745 and supports the benefit-risk analysis required by Annex I Chapter I. It is the data generated without human subjects that justifies whether the device is safe enough to be placed on the market or to enter a clinical investigation under MDR Article 61.
- Bench testing covers mechanical, electrical, chemical, software, performance, interoperability, stability, shelf life, and simulated-use studies — every test that answers a specification question without involving a patient. Each bench test belongs in the file as a controlled record with a method, an acceptance criterion, a result, and a traceable link.
- Animal studies are required only when bench data and in vitro data cannot answer a specific risk or performance question, and the scientific justification for running them has to be documented explicitly before the study starts, not retrofitted afterwards.
- The 3Rs — replace, reduce, refine — are not just an ethical framing. They are the default structure an auditor expects to see in the justification for any animal study referenced in the file.
- Good Laboratory Practice framing applies when the pre-clinical study is intended to support safety claims for regulatory submission. Not every bench test needs to be run under GLP, but safety studies intended as pivotal evidence typically should be, and the file should state clearly which regime each study was run under.

---

## Pre-clinical versus clinical — the distinction that decides where evidence lives

Pre-clinical and clinical data sit next to each other in section 6 of Annex II and they are constantly confused. They are not the same thing and a file that blurs them gets findings.

Pre-clinical data is every piece of evidence generated without a human subject. Bench testing on a hardware prototype. Tensile testing of a cannula. Bioburden measurement on a sterile device. Software performance testing on a reference dataset. Simulated-use testing on a phantom. Animal testing where it is scientifically justified. These are pre-clinical records and they sit in section 6.1.

Clinical data is evidence generated from the use of a device on human subjects. It comes from clinical investigations run under MDR Article 62 and the requirements of Annex XV, from post-market clinical follow-up, from scientific literature on the device or on equivalent devices, and from the clinical evaluation process described in MDR Article 61 and Annex XIV. These records sit in section 6.1 too, but clearly separated from the pre-clinical evidence and marked as clinical.

MDR Article 61(1) is explicit that confirmation of conformity with the relevant general safety and performance requirements shall be based on clinical data, and that the clinical evaluation shall follow a defined and methodologically sound procedure based on a critical evaluation of scientific literature, of the results of clinical investigations, or of both. Pre-clinical evidence does not replace clinical evaluation. It complements it. It answers the questions that can be answered without putting a device in a human, so that the clinical investigation — when one is required — can be designed around the residual questions that only human use can answer.

The practical consequence for the file is simple. Before an auditor can read the clinical evaluation report with confidence, the pre-clinical evidence needs to have already closed every question it is capable of closing. If the bench testing is incomplete and the team is trying to use the clinical investigation to cover the gap, the file is inverted and the clinical data will not carry the weight the team hopes it will.

## Bench test documentation — the workhorse of section 6.1

Bench testing is the backbone of pre-clinical evidence for almost every device. It is also the category where startups produce the most data and document it the worst.

A controlled bench test record has a fixed shape. It identifies the test item — exact device configuration, hardware revision, software version, lot where relevant. It states the hypothesis or the specification under test, referenced by identifier to the design inputs and to the GSPR row the test is feeding. It describes the method in enough detail that a third party could reproduce it, including equipment, calibration status of measuring instruments, sample size, environmental conditions, and acceptance criterion. It records the raw results, the processing applied, and the final pass or fail decision under an owner's signature or electronic equivalent. It carries a stable identifier, a version number, a date, and a status under the document control procedure required by the quality management system.

Typical bench test categories that land in section 6.1 include mechanical testing (tensile, compression, fatigue, burst, torsion), electrical testing against the relevant parts of the EN 60601 series where applicable, chemical and material testing including leachables and extractables where patient contact is involved, stability and shelf-life testing under defined storage conditions, sterility and bioburden testing for devices supplied sterile, packaging integrity testing, transport simulation testing, software performance testing on reference datasets, interoperability testing with connected devices or accessories, and simulated-use or phantom testing that exercises the device in a bench approximation of its intended use.

The mistake that gets flagged most often is the informal bench report. A test was run, the engineer wrote a memo, and the memo lives in a shared folder. The memo is not under document control, does not cite an acceptance criterion against the design inputs, does not identify the exact test item version, and does not carry a version of its own. Under an audit it produces a finding. The fix is not to rerun the test — it is to redo the report properly, under document control, citing the original data, with the acceptance criterion and the linkage made explicit.

The [verification and validation evidence post](/blog/verification-validation-evidence-technical-documentation) walks through the QMS framing that each bench test sits inside, and the [MDR Annex II structure post](/blog/mdr-annex-ii-structure) places section 6.1 in the full file.

## Animal study documentation — when it is needed and what the file has to show

Animal studies under MDR are the exception, not the rule. They are scientifically justified only when a specific safety or performance question cannot be answered by bench testing, in vitro testing, or computational modelling, and when the animal model provides meaningful predictive value for the human use case. The file has to document this justification explicitly before the study runs.

A pre-clinical animal study record in section 6.1 contains a set of items that go beyond what a bench report needs. It starts with the scientific rationale — the specific questions the study is designed to answer, why those questions cannot be answered by non-animal methods, and why the chosen species and model are predictive for the intended human use. It references the study protocol, which should have been approved by the relevant ethics committee or animal welfare body before the study began, under the applicable national implementation of EU Directive 2010/63/EU on the protection of animals used for scientific purposes.

The study report then covers the items an auditor expects to sample: the species, strain, age, sex distribution, and number of animals; the test and control groups and their randomisation and blinding where relevant; the surgical or interventional procedures; the device configurations used and their traceability to production-equivalent hardware; the observation period and endpoints; the histopathology, imaging, or analytical methods applied at termination; the statistical plan and its execution; the raw and processed data; the adverse events observed; and the study conclusions tied back to the original scientific questions.

The file also has to state the regulatory regime the study was run under. Was it conducted under Good Laboratory Practice at a certified facility, or was it conducted outside GLP with a documented quality plan? Was the facility inspected by a competent authority? Was the study run internally or by a contract research organisation, and if external, is the contract and the quality agreement on file? These are not paperwork questions. They are the items the auditor needs to decide how much weight to give the study in the benefit-risk analysis.

The implicit rule is that any animal study referenced in the file should stand as evidence on its own. The auditor should be able to read the study record and decide independently whether its conclusions are supported. A study whose report is a summary slide deck and whose underlying data lives somewhere else will not carry weight.

## The 3Rs — replace, reduce, refine — as a documentation requirement

The 3Rs principle — replace animal use where possible, reduce the number of animals to the scientific minimum, refine procedures to minimise suffering — is embedded in Directive 2010/63/EU and is the framework any ethics committee will expect to see in the study justification. It is also increasingly the framework notified bodies expect to see reflected in the pre-clinical documentation.

A 3Rs justification in the file reads as three explicit arguments. Replace: the team considered non-animal alternatives — bench tests, in vitro tests, computational models, existing literature on similar devices or materials — and documented why each alternative was insufficient to answer the specific question at hand. Reduce: the sample size calculation is documented, using the minimum number of animals consistent with statistical validity, and the study design exploits every legitimate technique for extracting more information per animal. Refine: the procedures, anaesthesia, analgesia, housing, and humane endpoints are defined in the protocol, consistent with current best practice, and the study is run by personnel competent in the techniques involved.

A file that cites an animal study without a 3Rs justification looks careless. A file that reflects the 3Rs analysis explicitly looks like the work of a team that understands why the study was run and what its limits are. This is a readability signal as much as a compliance signal.

## GLP framing — when it applies, when it does not, and how to state it in the file

Good Laboratory Practice is the quality system developed for non-clinical safety studies intended to support regulatory submissions. It is not, in itself, required by MDR for every bench test. A mechanical pull test on a catheter does not need to be run under GLP to be admissible evidence. A long-duration safety study in an animal model that is intended as pivotal support for a safety claim often does.

The decision about which studies to run under GLP is a benefit-risk-analysis decision informed by the role the study plays in the overall conformity argument. Studies that are the principal evidence for a safety endpoint, that will not be repeated, and that support a claim the notified body is expected to scrutinise are the strongest candidates for a GLP framing. Routine design verification bench tests run inside the QMS are usually not.

What the file has to do in all cases is state clearly which regime each study was run under, so the auditor is not left to guess. For each pre-clinical study in section 6.1, the record should identify whether the study was conducted under GLP, under the manufacturer's QMS, or under another defined quality regime, and it should name the facility where the work was done. A study whose quality regime is not stated is a study whose credibility the auditor cannot assess.

## How pre-clinical evidence integrates with the risk file and the clinical evaluation

Pre-clinical evidence is not a standalone chapter. It is the load-bearing link between two other parts of the file that are frequently read on their own.

The first link is into the risk management file under EN ISO 14971:2019+A11:2021. Every hazard identified by the risk analysis that is addressed by a design control or a protective measure needs evidence that the control works. Much of that evidence is pre-clinical. A mechanical hazard is controlled by a design feature whose effectiveness is demonstrated by a bench test. A biocompatibility hazard is addressed by material choice and by the biological evaluation under EN ISO 10993-1:2025. A software hazard is addressed by a control whose effectiveness is demonstrated by software verification. The risk file names the control; the pre-clinical evidence proves it works; the link between them is the stable identifier. A pre-clinical study without a named risk it closes, and a risk control without a named piece of pre-clinical evidence confirming it, are both incomplete halves of the same argument.

The second link is into the clinical evaluation under MDR Article 61 and Annex XIV. The clinical evaluation report states which clinical claims need clinical evidence and which questions about safety and performance can be answered by pre-clinical data. Where pre-clinical data closes a question, the clinical evaluation cites the pre-clinical record by identifier and does not duplicate it. Where pre-clinical data is insufficient, the clinical evaluation has to identify that gap and explain how it will be closed — typically by clinical investigation or by post-market clinical follow-up. This traffic between sections is where the auditor decides whether the benefit-risk analysis holds together.

The [clinical evaluation cluster posts](/blog/clinical-evaluation-report-mdr) and the [benefit-risk analysis post](/blog/benefit-risk-analysis-technical-documentation) go deeper on the downstream integration.

## Common pre-clinical documentation gaps

The failure patterns in startup section 6.1 repeat across files.

- **Bench tests run without a controlled report.** The data exists in a lab notebook or a CI artifact. No document-controlled record ever gets written. At audit the team discovers that raw data without a controlled report is not evidence.
- **No linkage to design inputs or GSPR rows.** The test was real, the method was sound, but nothing in the record identifies which requirement the test was proving. The auditor cannot use it.
- **Animal study referenced with no 3Rs justification.** The study was run years ago, possibly under a previous company, and the justification for running animals was never written down. The file now has to reconstruct a justification that should have been prospective.
- **GLP status not stated.** The study was either GLP or it was not, and the record does not say. The reader defaults to assuming it was not, which often weakens the evidence unnecessarily.
- **Stability and shelf-life testing deferred until after submission.** The labelled shelf life is claimed in the IFU, but no stability study supports it yet. This is a common gap and it blocks submissions.
- **Bench reports for superseded prototypes left in section 6.1.** The test was run on an earlier revision. The production device differs in ways that matter. The file still cites the old test as current evidence.
- **Simulated-use testing skipped in favour of clinical investigation.** The team skips bench simulations and hopes the clinical investigation will cover everything. The clinical investigation then gets findings the team cannot answer because the pre-clinical work was not done.
- **Biological evaluation run once and never revisited.** Material suppliers change, processing changes, sterilisation changes. The biological evaluation under EN ISO 10993-1:2025 is not re-assessed and the file drifts.

Every one of these is preventable at the moment the work is done. The cost of doing the record properly up front is a small fraction of the cost of reconstructing it under audit pressure.

## The Subtract to Ship angle

Pre-clinical files attract the same accumulation problem as the rest of section 6. Engineers run exploratory tests, parallel tests, early-prototype tests, and edge-case studies, and every one of those records feels valuable to the person who produced it. The live pre-clinical evidence chain only needs the records that trace to a design input, a GSPR row, or a risk control in the current design. Everything else is engineering memory, not evidence.

The [Subtract to Ship framework for MDR](/blog/subtract-to-ship-framework-mdr) applies to section 6.1 the same way it applies elsewhere: every record in the live file earns its place by tracing to something, and records that do not trace come out — either deleted, or moved out of section 6.1 into an engineering archive where they can inform future work without muddying the current conformity argument.

A disciplined pass on a typical startup section 6.1 removes the superseded-prototype reports, the exploratory studies that reached no conclusion, the parallel test series that were not selected as the canonical evidence, and the informal memos that were never meant to be controlled records in the first place. What remains is smaller, faster to audit, and internally consistent. The file does not need to be heavy. It needs to be right.

## Reality Check — Where do you stand?

1. For every bench test in your section 6.1, can you name the specific design input or GSPR row the test was answering and the acceptance criterion it was tested against?
2. Is each pre-clinical study in your file a controlled document with a stable identifier, a version, an owner, and a status — or are some of them lab memos that have never been placed under document control?
3. For every animal study referenced in your file, is there a prospective 3Rs justification explaining why non-animal methods were insufficient, how the sample size was minimised, and how the procedures were refined?
4. Does each pre-clinical study record state explicitly which quality regime it was run under — GLP, the manufacturer's QMS, or another named regime — and at which facility?
5. For every risk control in your risk management file that relies on pre-clinical evidence, can you point to the specific study in section 6.1 that verifies the control works?
6. Is your biological evaluation under EN ISO 10993-1:2025 current with the materials, processing, and sterilisation used in the production device, or was it done once on an earlier revision and never revisited?
7. Does your stability and shelf-life evidence support the shelf life claimed on the label, or is the claim outrunning the data?
8. Have superseded-prototype test reports been removed from your live section 6.1, or do they still sit next to current reports without a clear indication of which is which?

## Frequently Asked Questions

**What is pre-clinical testing documentation under MDR?**
Pre-clinical testing documentation under MDR is the set of controlled records in Annex II section 6.1 of Regulation (EU) 2017/745 that document every piece of non-clinical evidence generated to support the benefit-risk analysis of the device. It includes bench tests, in vitro studies, simulated-use testing, software performance testing, and, where justified, animal studies. It is the evidence that answers every safety and performance question that can be answered without involving a human subject.

**Where in the technical documentation does pre-clinical evidence live?**
It lives in Annex II section 6.1 of Regulation (EU) 2017/745, alongside — but clearly separated from — the clinical evidence. The same section holds the clinical evaluation report, but the pre-clinical content is labelled and organised so an auditor can read each category on its own.

**Are animal studies always required for MDR certification?**
No. Animal studies are required only when the specific safety or performance question cannot be answered by bench testing, in vitro methods, computational modelling, or existing literature, and when the animal model provides predictive value for the human use case. The justification has to be documented prospectively, before the study is run, and has to be consistent with the 3Rs principle of replace, reduce, refine.

**Do bench tests need to be run under Good Laboratory Practice?**
Not all of them. Routine design verification bench tests run inside the manufacturer's QMS do not need GLP. Pivotal non-clinical safety studies intended as principal evidence for a safety claim in a regulatory submission often should be run under GLP, or under a clearly documented equivalent quality regime. The file has to state explicitly which regime each study was run under.

**How does pre-clinical evidence connect to the clinical evaluation under Article 61?**
The pre-clinical evidence closes every question it is capable of closing, and the clinical evaluation under MDR Article 61 and Annex XIV then addresses the residual questions that only human use can answer. The clinical evaluation cites the pre-clinical records by stable identifier rather than duplicating them, and where pre-clinical data is insufficient, the clinical evaluation has to identify the gap and say how it will be closed — typically by clinical investigation or by post-market clinical follow-up.

**How does pre-clinical evidence link to the risk management file?**
Under EN ISO 14971:2019+A11:2021, every risk control has to be verified. For many hazards — mechanical, electrical, chemical, biological, software — the verification evidence is pre-clinical. The risk file names the control, the pre-clinical record in section 6.1 proves the control works, and the two are linked by stable identifiers so an auditor can walk from a hazard to its evidence in either direction.

**What is the 3Rs principle and why does it matter for the file?**
The 3Rs — replace, reduce, refine — is the ethical and scientific framework for animal research embedded in EU Directive 2010/63/EU. For the file, it means any animal study referenced has to carry an explicit prospective justification showing that non-animal methods were considered and found insufficient (replace), that the sample size was the minimum consistent with statistical validity (reduce), and that the procedures and welfare measures were aligned with current best practice (refine). A file that cites animal studies without this justification reads as careless.

## Related reading

- [Technical Documentation Under MDR: What It Is and Why Startups Get It Wrong](/blog/technical-documentation-under-mdr) — the hub post for the technical documentation cluster this pre-clinical post sits inside.
- [Pre-Clinical Data and Engineering Studies in the Technical File](/blog/preclinical-data-engineering-studies-technical-file) — the adjacent post on how pre-clinical and engineering studies are organised inside section 6.
- [How to Document General Safety and Performance Requirements (Annex I Checklist)](/blog/document-gspr-annex-i-checklist) — the section 4 checklist each pre-clinical record has to trace to.
- [Verification and Validation Evidence in Technical Documentation](/blog/verification-validation-evidence-technical-documentation) — the QMS framing for design verification and validation the bench tests sit inside.
- [Clinical Evaluation Plan Requirements Under MDR Article 61](/blog/clinical-evaluation-plan-mdr-article-61) — the downstream document that consumes pre-clinical evidence when building the clinical argument.
- [Benefit-Risk Analysis in the Technical Documentation](/blog/benefit-risk-analysis-technical-documentation) — the section 5 content the pre-clinical evidence has to support.
- [Biocompatibility Evidence Under EN ISO 10993-1](/blog/biocompatibility-evidence-iso-10993) — the biological evaluation record that sits inside the pre-clinical chapter for any device with patient contact.
- [Stability, Shelf Life, and Aging Studies in the Technical File](/blog/stability-shelf-life-aging-studies) — the specific pre-clinical category that supports labelled shelf life claims.
- [The Subtract to Ship Framework for MDR Compliance](/blog/subtract-to-ship-framework-mdr) — the methodology behind a lean, disciplined pre-clinical file.

## Sources

1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Annex I Chapter I (general safety and performance requirements, including the benefit-risk analysis), Annex II section 6 (pre-clinical and clinical data), Article 61 (clinical evaluation). Official Journal L 117, 5.5.2017.
2. EN ISO 14971:2019 + A11:2021 — Medical devices — Application of risk management to medical devices.
3. EN ISO 10993-1:2025 — Biological evaluation of medical devices — Part 1: Evaluation and testing within a risk management process.
4. EN ISO 14155:2020 + A11:2024 — Clinical investigation of medical devices for human subjects — Good clinical practice.
5. Directive 2010/63/EU of the European Parliament and of the Council of 22 September 2010 on the protection of animals used for scientific purposes. Official Journal L 276, 20.10.2010.

---

*This post is part of the Technical Documentation & Labeling series in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. Tibor has reviewed pre-clinical chapters as a notified body lead auditor and assembled them as a MedTech founder, and the pattern is the same on both sides of the table: the files that pass cleanly are the ones where every bench test and every animal study was set up, from day one, to answer a specific question that someone else in the file was also asking.*

---

*This post is part of the [Technical Documentation & Labeling](https://zechmeister-solutions.com/en/blog/category/technical-documentation) cluster in the [Subtract to Ship: MDR Blog](https://zechmeister-solutions.com/en/blog). For EU MDR certification consulting, see [zechmeister-solutions.com](https://zechmeister-solutions.com).*