Table of contents
At a glance
How do you select the appropriate GAMP category?
GAMP 5 places infrastructure software, tools and IT services in category 1, standard components in category 3, configured components in category 4, and custom-developed applications or components in category 5. GAMP 5 removed the former firmware category 2 in 2008. Assess each relevant component as implemented. The category helps planning but determines neither GxP risk nor the validation scope on its own. One custom script does not make every other component custom software.
- Categorize identifiable components instead of trade names.
- Separate specification, GxP implications, and available evidence.
- Handle edge cases based on the actually modified logic.
- Derate evidence from specific risks and open questions.
1. From product designation to the specific decision
GAMP is a methodological guide of the ISPE and not a legal classification of software products. In the human medicine GMP context, the actual use remains to be assessed under the applicable requirements. Annex 11 focuses among other things on system function, risks and evidence. A category number does not replace this consideration.
The 2026 ISPE technical article outlines categorization primarily at the component level. An early holistic view can assist in planning supplier assessment; however, this does not result in a complete validation checklist. The value lies in providing a clear understanding of a component's characteristics.
For practical implementation, the work begins with a question: Which function are you actually modifying or using here? Only after answering this question should category, rationale, and required evidence be considered. A sales statement may serve as a hint, but it does not alone address this question for your specific installation.
2. Understand the categories 1, 3, 4, and 5 along with their intended use
The known numbers denote different types of software components. The following concise overview links the historical numbering to the component types described in the public ISPE contribution of 2026. This is not a reproduction of the paid guide nor a list of prescribed documents.
Category 2 previously designated firmware as a separate software category. The ISPE report on the implementation of GAMP 5 from 2008 already addressed this distinction. Therefore, firmware is not a missing regulatory category and was not eliminated solely with the 2022 second edition.
- 1: Infrastructure software, tools, and technical services as the foundation of application usage.
- 3: Standard components without customer-specific configuration of the business process; pure runtime parameters may still be present.
- 4: Components whose existing functions are configured for the specific business process.
- 5: Applications or components featuring custom-developed or adapted program logic.
3. Prepare the correct classification with accurate information provided
First, draw a small data flow: where do the information originate, which component modifies it, and what decision uses it? Assign version, configuration, custom extensions, and operator responsibility accordingly. Separate only to the extent that differences become clearly understandable for audit and change documentation purposes.
Ask the supplier specifically about the boundary between delivered functionality and your customizations. A configuration extract is more helpful than a general response of "everything standard". For any custom code, you need a clearly identifiable version and a description of its intended function. Any unconfirmed components remain visible as open questions.
A useful classification also indicates the information it contains. This enables a subsequent reviewer to determine whether the conclusion is based on a widely accepted configuration, a manufacturer's specification, or merely on an assumption still awaiting validation. A provisional entry is more appropriate than an ungrounded final determination.
4. Complete Example: A quality status is carried forward
An operation manages the quality status of internal material bins. A standard module displays data records, a configurable workflow enables the intended status transition, and a dedicated script transfers selected information to a planning list. This list may include, in the example, only records with status "ready".
The considered state is referred to as F4. Its six components are fully assigned below for this specific case. "Fully" refers to the described example, not to any conceivable enterprise architecture. Physical devices and additional operational services would need to be assessed separately in your project, depending on their impact.
The last column contains proposed decisions for validation conclusions. These are not prescribed checklists for any specific category. Each row links attribute characteristics to specific failure scenarios, thereby transforming the categorization into a actionable task.
| Component and Function | Classification with rationale | Specific impact in the example | Decision on evidence |
|---|---|---|---|
| B7: Operating system and runtime | 1; technical basis, no individual professional logic described | System failure or inappropriate status may impact application and export. | Confirm required version and operating basis; consider retry following actual process need |
| D3: Standard Database Service | 1; Storage as a technical platform; separately evaluate professional tables/application logic | Missing or incorrectly restored data alters the working status. | Assessment of security and recoverability of relevant data, as well as interoperability with F4 |
| V1: unaltered dataset display | 3; shipped notice without an independent business process; display parameters remain standard parameters | Users might see the wrong packaging or an incomplete status. | Take supplier documentation for approval; verify local assignment and required fields |
| W4: configured status workflow | 4; existing states and transition rules are suspended for the procedure flow | Unauthorized transition could result in a locked batch being marked as ready. | Conduct configuration review together with relevant role, transition, and failure scenarios |
| E2: In-house developed export | 5; own code selects datasets and forms the transfer | Incorrect selection or assignment could result in a blocked batch being released. | Regulatory framework and requirements are integrated; targeted verification of selection, identification, status, and error reporting |
| J1: Standard time control | 1; technical tool initiates E2 without inherent subject-matter selection rules | Duplicate or missing executions may compromise the planning list. | Consider start-up conditions and re-start; review double processing in conjunction with E2. |
5. Translate the matrix into a verifiable decision
In the test example, three data records are prepared: G-31 is marked as "ready", G-32 as "locked", and G-33 as "under review". The predefined expected behavior for E2 is that only G-31 appears in the planning list with unchanged identifier. The other two records remain excluded. The category number alone does not fulfill this expected outcome.
Assuming a change in status labels also transfers E2 to G-32, the relationship between W4, data transfer, and export is then investigated. A script correction requires appropriate repeat and follow-up validations. However, this does not imply that the unchanged operating system B7 would now require individual development.
A second case involves time-based control: if the same run is repeated, the handling of previously transferred containers as required in the example must be verified. The platform component may remain in Category 1, despite its critical interaction with E2 for process integrity.
The decision does not amount to "Category 5: test everything anew." It specifically identifies affected functions, changed prerequisites, appropriate existing evidence, and concrete gaps. This ensures the scope is clearly justified and prevents overlooking any impacts outside the script.
6. Clarify edge cases between parameters, configuration, and development
The critical boundary does not align along a user interface. A graphical editor can generate complex process logic; text input can only set a runtime parameter unchanged. Investigate the impact of the setting and determine the actual logic that is generated.
The following work aid deliberately does not make any decisions solely based on keywords such as No-Code, Standard, or Interface. It identifies the information still required for a justified assignment. The example scenarios are independent of the previously described status F4.
| Situation | Key question | Justified Treatment |
|---|---|---|
| Standard display with configurable time zone | Is only the existing display being parametrized or is business decision logic being altered? | Pure runtime parameter does not automatically turn the component into Category 4; check its effect anyway |
| Workflow in the graphical editor | Are existing states/rules configured or is custom code embedded? | Consider configured workflows; identify individual code components separately. |
| Standard connector with field mapping | Does he transfer unchanged, modify the configuration, or invoke his own code? | Standard function, mapping, and functional expansion to be logically separated; data impact determines validation questions |
| Small macro in widely used standard software | Which own logic processes which decision-relevant data? | Evaluate macro as an individual component; do not generically reclassify the base software. |
| Well-maintained extension by the supplier | Is it provided unchanged, configured for your use, or custom-developed? | Supplier origin alone does not determine; specific implementation and available evidence are assessed. |
7. Evaluate category, GxP implications, and uncertainty separately
A category number does not indicate the severity of the consequences resulting from a failure. In the example, an incorrect status display for a standard component could lead to an inappropriate material selection. In contrast, a single export for purely informal analysis might have a different impact. The actual usage determines these consequences.
The public ISPE contribution of 2026 highlights complexity and novelty. It indicates that a complexly configured component may warrant greater scrutiny than a simple individual component. However, this does not establish a new general reversal rule.
For each key function, document what could fail, which control detects or prevents the failure, and what uncertainty remains. ICH Q9(R1) supports a risk- and knowledge-base adjusted evaluation. A artificial rule such as "category times criticality equals test frequency" does not adequately explain these technical relationships.
8. Link supplier evidence and changes to the relevant components
When inquiring with a supplier package, please specify which exact statement is supported for which module component. A test of the standard display version V1 may be useful, while the selection rule in E2 or your workflow W4 falls outside the scope of the tested scope. Conversely, a supplier may provide appropriate documentation for a custom-developed extension.
Therefore, in addition to the classification, provide evidence of scope. A category is neither a quality certificate nor a justification for disregarding appropriate documentation. What remains decisive is whether the status, conditions, and results address your question.
Upon modification, re-evaluate the characteristics and performance. When a simple field mapping is replaced by a self-programmed transformation, the rationale for this component changes. Even if the category remains unchanged, a different technical justification may be required. A new blocking rule in W4 is already sufficient for this purpose in the example.
9. Address the classification within the SME with clearly defined responsibilities
For an initial joint walkthrough, you will require process knowledge, awareness of the actual configuration in use, and access to relevant supplier documentation. Not every participant needs to be proficient in all technical details. However, identify who is responsible for answering open questions and who is capable of evaluating the resulting planning.
First address components whose failures could impact critical decisions or whose implementation remains unclear. Flag missing information as an open task with the responsible party. A complex edge case must not result in the remaining well-understood sections being left unaddressed for months.
The effort involves, among other factors, interfaces, in-house developments, configuration complexity, and data gaps. The number of categories provides little insight. Plan based on specific tasks such as configuration review, data reconciliation, or code review; blanket time and cost estimates would be unfeasible without this information.
10. Checklist for a robust component selection decision
This internal support document assists in completing your evaluation. It is not an ISPE form. A different qualified reviewer should understand the decision based on the provided information and be able to address specific changes as needed.
If a mapping remains unclear, first describe the open technical question and the required safeguarding. An uncertain value must neither replace a relevant test nor automatically trigger a complete maximum package.
- The intended use and its GxP boundary are defined.
- Components can be distinguished by function and actual current status.
- Standard function, process configuration and own code are not mixed.
- The rationale is based on verifiable information.
- Unclear assumptions are visible and assigned as a clarification task.
- The GxP impact is assessed independently of the category number.
- Relevant interdependencies between components have been identified.
- Supplier invoices are assigned with their specific scope of coverage.
- Decision on substantiation addresses a risk or an open question.
- Changes can be traced back by version, configuration, and decision.
Project experience and collaboration
Assess the software components behind the product name
An application can combine standard functions, configuration and custom code. With IT, process owners and, where relevant, the supplier, we map the components and identify customer-specific logic. This helps distinguish usable existing evidence from local testing needs. A small extension may warrant assessment limited to its function and dependencies. A new platform needs a broader inventory first. A category number alone establishes neither a risk level nor a test plan.
Frequently asked questions about GAMP software categories
Can a system contain multiple GAMP categories?
Yes. An application can integrate technical platforms, unaltered standard functions, configured workflows, and custom code. Describe these components in such a way that their differences are actionable for validation and change management purposes. A high-level overall assessment may assist in early supplier inquiries; it should not replace subsequent component-level evaluation and professional risk assessment.
Sources: [1]
Why is there no separate software category 2 in GAMP 5?
Category 2 previously referred to firmware. The public ISPE report from May/June 2008 already identified the removal of this standalone category during the transition to GAMP 5. This did not occur only in 2022. The removal of firmware from consideration does not automatically render it non-critical or exempt from validation evidence. Its specific function and characteristics must still be understood within the relevant system.
Is Category 5 always more risky than Category 3?
No. The number is not an indication of the priority or likelihood of a GxP-related event. An unaltered standard function can still support a critical decision. Complexity, novelty, effective controls, and available evidence all influence the assessment. Evaluate the specific failure scenario. Neither a low number nor the designation "standard" alone justifies a reduced level of validation evidence.
Does a small script make the entire system fall under Category 5?
Not in a flat. Identify the individual component, its task and its dependencies. In the example of this article, E2 is its own code, while B7 retains its role as a technical basis. However, the required check area may go beyond the script, for example if its output influences a subsequent process. Limited categorization therefore does not mean isolated testing without handovers.
Is standard software with settings always classified as Category 3?
Not every setting is identical. A time zone or runtime parameter may remain part of an unchanged standard function. When existing functions are combined to align with a specific business process logic, the resulting configured component must be evaluated differently. Describe the actual impact of the setting. The mere presence of a settings menu or its label does not define the boundary case.
Sources: [1]
Are No-Code Workflows inherently classified as Category 4?
The technical implementation must be evaluated. A graphically configured workflow can combine existing functions; embedded scripts or custom components can enhance specific logic. Separate these elements rather than assigning a platform designation. Even in the absence of proprietary code, a complex workflow may require targeted validation evidence. The user interface provides limited insight into its functional impact.
Can we adopt the supplier's category designation?
Use them as justifiable information. Inquire about the affected product lots, considered components, and excluded modifications. A plausible alignment of the standard core may apply, while local configurations and enhancements should be evaluated separately. Similarly, appropriate supplier documentation can support your assessment. Neither uncritical acceptance nor blanket rejection aids in the specific evidence planning.
Which documents are strictly required for each GAMP category?
A category number does not inherently define a universal document package. Define which requirements, decisions, and outcomes must be traceable for the specific intended use. From this, appropriate evidence and its organization follow. The public ISPE contribution of 2026 explicitly cautions against a purely checklist-based approach. Internal procedures and applicable regulatory requirements must also be considered.
Must we reclassify after every change?
First assess whether the issue involves availability, functionality, or dependencies. A mere version change does not necessitate a new category; the introduction of new, standalone program logic may alter the classification of a component. Document the decision and its rationale in relation to the current status. Even if the category remains unchanged, additional testing may be required if the technical behavior has changed.
Is a low GAMP category a compliance demonstration?
No. It describes a methodological classification, not a regulatory approval or validation of your installation. Whether the system meets its intended use depends on appropriate requirements, controlled implementation, sufficient validation evidence, and proficient operation. A low-numbered classification neither masks open defects nor replaces missing evidence. The same applies in reverse for a comprehensive documentation set belonging to a higher category.
Sources and review date
Source date: 6 October 2026. Public ISPE contributions and publisher overviews were reviewed; the paid GAMP full text was not accessed. Contributions from 2008/2010 confirm historical categorization and numbering. Component matrix, edge cases, and test cases are standalone reference tools. Regulatory focus: Human medicinal product GMP. Targeted source cross-check conducted on 18 October 2026: ISPE technical article "When Is a Category Not a Category?" (January/February 2026) regarding component designations in the second edition, as well as the ISPE Update May/June 2008 on the prior firmware category.
- ISPE / Karen Ashworth: When Is a Category Not a Category?
- ISPE: GAMP 5 Guide, 2nd Edition – Public Publisher Overview
- ISPE Update: GAMP 5 Debuts to a Record Breaking European Audience
- ISPE: Guide for using Open Source Software in Regulated Industries based on GAMP
- European Commission: EU GMP Annex 11 – Computerised Systems
- ICH Q9(R1): Quality Risk Management – official EU version
Sources checked: .
From category to specific evidence questions
We support you in documenting software components, configurations, and custom extensions in a traceable manner. From this, we work with your team to develop a justified plan for the required validation evidence.
- GAMP 5 consulting →
Understand how GAMP 5, EU GMP Annex 11 and 21 CFR Part 11 apply to your systems. We translate the relevant requirements into proportionate validation work and evidence your team can defend in an audit.

