
On this page
Published: · Updated:
Key Takeaways
- Use eSTAR 7.1 for current-version planning and confirm the applicable template with the FDA before filing.
- Section 524B duties, guidance recommendations, and attachment packaging choices are different.
- Map applicable prompts to evidence titles, revisions, and sections, not a universal number of files.
- Architecture, threats, SBOM components, controls, and security testing should describe the same configuration.
- Technical screening checks completeness; substantive review evaluates device-specific evidence.
How should manufacturers prepare eSTAR 7.1 cybersecurity documentation?
Prepare eSTAR 7.1 cybersecurity documentation against the applicable template prompts, Section 524B duties, and device-specific evidence recommendations. Map each prompt to an attachment and section. Keep architecture, threat modeling, risk, SBOM, testing, labeling, and postmarket records consistent. Document counts and attachment groupings are organizational choices, not universal FDA requirements.
Patient safety depends on security controls that work in the device's intended environment. This guide combines attachment mapping with document-by-document preparation checks. It is for manufacturers, engineering teams, and quality and regulatory leads preparing cybersecurity evidence.
The primary source is the FDA's Cybersecurity in Medical Devices: Quality Management System Considerations and Content of Premarket Submissions, issued February 3, 2026. Section 524B adds statutory duties for applicable cyber devices. Neither the guidance nor Blue Goat's organizational checklist creates a universal required number of separate reports.
The document examples below come from published FDA recommendations, scenarios in that guidance, and a public SPDX sample. They are not completed client reports. Populate your records from actual engineering evidence; no example establishes a device's risk or test result.
Start with the applicable eSTAR 7.1 template
Download the appropriate IVD or non-IVD template from the FDA eSTAR Program. Record the version and device configuration before mapping evidence. Use eSTAR 7.1 for current-version planning, and confirm the FDA's instructions for your pathway and any response to an existing submission.
Covered 510(k) submissions and covered De Novo requests use eSTAR, subject to applicable exemptions. Other pathways have their own instructions. The form shows or hides prompts based on your answers, so a static attachment list cannot replace the actual template.
Moving an older draft into eSTAR 7.1
- Record the old template version and the revision of every attachment.
- Open the applicable eSTAR 7.1 template and compare its prompts with the prior draft. Do not assume section numbers or field labels are unchanged.
- Reconcile controls descriptions and software, risk, labeling, and performance references against the device evidence.
- Carry forward only records that still describe the submitted build. Reassess changed interfaces, components, assumptions, and tests.
- Follow the FDA's current transition and response instructions for already-filed submissions. A template revision alone does not establish that a device requires a new submission.
This is a migration review checklist, not a verified field-by-field difference between template versions. The current template and its version history are the source for exact changes.
Does Your Device Qualify as a Cyber Device Under Section 524B?
A cyber device must meet all three statutory criteria, not merely contain a wireless interface. It includes software validated, installed, or authorized by the sponsor, has the ability to connect to the internet, and contains technological characteristics validated, installed, or authorized by the sponsor that could be vulnerable to cybersecurity threats.
Assess direct and indirect internet capability in the intended device configuration. Document the rationale rather than classifying the device from one interface alone. See Section 524B Cybersecurity Requirements Explained for the statutory decision points.
For applicable cyber devices, Section 524B addresses vulnerability monitoring and management, processes to provide reasonable assurance of cybersecurity, updates and patches, and an SBOM. The statute does not establish a universal number of PDF reports. Devices outside that definition may still need cybersecurity evidence appropriate to their risks and applicable submission prompts.
For a modification, document the cybersecurity impact analysis. Trace the change through architecture, components, threats, controls, and testing before determining the applicable evidence. A statement that no security features changed does not replace that analysis.
How eSTAR 7.1 maps cybersecurity evidence
Use the current applicable template to map evidence to prompts. In the eSTAR 7.1 map, several deliverables share a field; file splitting is a packaging decision, not proof that the evidence is complete.
Respond to each applicable prompt and identify the attachment and section containing its supporting evidence. Do not describe a 14-item or 18-item organizational checklist as a universal FDA-mandated document count.
| Evidence group | What to check before uploading | Existing detailed reference |
|---|---|---|
| Management plan and SPDF evidence | Lifecycle responsibilities, vulnerability monitoring, and update processes are documented. | eSTAR 7.1 evidence mapping |
| Risk management and risk assessment | Threats, patient-harm scenarios, controls, and residual-risk rationale are connected. | Deliverables and eSTAR Map |
| Threat model and architecture | Interfaces, trust boundaries, architecture views, and threat scenarios describe the same system. | eSTAR 7.1 evidence mapping |
| SBOM and component evidence | Inventory, support information, vulnerability analysis, and scope assumptions can be cross-referenced. | SBOM Format Requirements |
| Controls and testing | Implemented controls link to test evidence, findings, remediation, and retests. | Deliverables and eSTAR Map |
| Metrics and unresolved anomalies | Applicable metrics and cybersecurity impacts of unresolved anomalies are addressed. | Unresolved Anomalies in FDA Cybersecurity |
| Labeling | Support information, user responsibilities, and security assumptions match the technical evidence. | eSTAR 7.1 evidence mapping |
Use these groups as a review aid, not as exact field labels for every template version. For field-by-field placement, consult the FDA Premarket Cybersecurity Deliverables & eSTAR v7.1 Map alongside the template you are submitting.
Evidence that deserves an explicit check
Address cybersecurity metrics according to the applicable prompts and available device history. If the device has no fielded-product history, state that context and distinguish planned monitoring from actual measurements. Never invent monitoring data to populate a report.
Assess unresolved software anomalies for cybersecurity impact. Link relevant anomalies to threat scenarios, controls, testing, and residual-risk rationale. Do not substitute an unassessed bug list or claim these reports are the most frequent cause of deficiencies without supporting evidence.
Applicability rationale and attachment index
Check: Record the device configuration, applicable template version, Section 524B determination, and each prompt's evidence location. Keep the statutory assessment separate from your attachment index.
Source example: Section 524B uses three criteria: sponsor-validated, installed, or authorized software; internet connectivity capability; and technological characteristics that could be vulnerable to cybersecurity threats. A wireless-interface label alone does not answer those questions. The FDA eSTAR Program provides the actual interactive template, not a universal list of filenames.
Document layout: Prompt text | applicability rationale | document title and revision | section/page | evidence owner | open gap. Copy the prompt from the template in use. A justified non-applicability entry needs its rationale, not merely an empty attachment cell.
Cross-check: The device description, connectivity assessment, architecture, and supplied software configuration should support the same determination.
Building Your Core Documentation: Threat Model, Risk Assessment, and SPDF Evidence
The threat model should represent the device system and inform security decisions. Cover relevant hardware, firmware, networks, wireless interfaces, cloud services, companion applications, users, and update infrastructure. Explain the chosen method, such as STRIDE or attack trees, and its scope.
Architecture views and their purpose
- Global system view: Identify the device ecosystem, external systems, interfaces, users, and communication paths.
- Multi-patient harm view: Describe shared dependencies and compromise paths that could affect multiple patients.
- Updateability and patchability view: Show update delivery, authorization, integrity checks, installation, and recovery behavior.
- Security use case views: Explain security-relevant operational scenarios, such as authentication, authorization, and failure handling.
Keep these views consistent with the software architecture. They may be organized together or cross-referenced clearly according to the applicable prompts. See eSTAR Software Documents and Cybersecurity: How They Connect for cross-section consistency checks.
Risk assessment should connect exploitability and security impact to device-specific patient-harm scenarios. CVSS can describe technical severity; it does not establish clinical risk by itself. The manufacturer owns residual-risk acceptance and the rationale supporting that decision.
SPDF evidence should demonstrate that security activities occurred throughout development. Include the relevant process records, requirements, design decisions, testing, and change-control evidence. Our SPDF Cybersecurity Documentation article explains how those records connect.
SPDF summary and lifecycle records
Check: Explain how security requirements, design reviews, threat modeling, testing, change control, and vulnerability handling operate within the manufacturer's development process.
Source example: Section V.C of the February 2026 guidance recommends cybersecurity testing throughout the SPDF. A lifecycle summary therefore needs links to executed security activities, not only a policy stating that testing will occur.
Document layout: Lifecycle activity | governing procedure | responsible role | actual review or test record | configuration/revision | resulting action. Use your dated records to show when a security requirement was reviewed, tested, and changed. Do not describe a planned review as an executed one.
Cross-check: Procedure references, design-review decisions, and test reports should identify the relevant software release. IEC 81001-5-1 can inform lifecycle activities; citing it does not establish certification or completed work.
Security risk management plan and report
Check: Separate the intended method from the report of what was actually done. Define evaluation criteria, roles, review triggers, and the relationship to the manufacturer's safety risk file.
Source example: Section V.A of the guidance recommends a security risk management report summarizing risk-evaluation methods, mitigation activities, residual-risk conclusions, and traceability to threat modeling, risk assessment, SBOM, and testing. A document containing only a scoring legend does not show those outputs.
Document layout: The plan identifies the method, criteria, decision authority, and update triggers. The report identifies the reviewed configuration, analysis records, implemented mitigations, remaining issues, and manufacturer decisions. Reference AAMI TIR57 and ANSI/AAMI SW96 where those methods are used.
Cross-check: A report's residual-risk conclusion must follow the device evidence and manufacturer review. Do not copy a risk-acceptance statement from another product.
Global system architecture document
Check: Diagram the device, internal and external connections, assets, interfaces, and trust boundaries. Pair the diagram with a flow description.
Source example: Section V.B.2(a) names software update infrastructure, intermediary connections, cloud connections, and patient home networks as elements to consider for interconnected devices. Appendix 2 asks for communication details, credentials, access controls, and links to hazards, controls, tests, and SBOM components.
Document layout: Diagram element | actual component | interface and direction | transmitted data or command | trust boundary | security control | SBOM reference. Include only elements present in the assessed system. A cloud symbol cannot replace the endpoint, authentication, and data-flow description.
Cross-check: Every in-scope interface in testing should be locatable in the architecture; exclusions need a recorded explanation.
Multi-patient harm architecture document
Check: Describe shared dependencies and compromise paths that could affect several patients, where applicable.
Published FDA scenario: Section V.B.2(b) describes multiparameter bedside monitors restarting simultaneously, leaving a shared network's monitors unable to monitor patient vitals while staffing cannot cover all patients. This is the FDA's illustrative scenario, not a Blue Goat finding or client incident.
Document layout: Shared asset or dependency | compromise path | affected device functions | potential multi-patient consequence | containment or recovery design | supporting test. Use the scenario to ask whether shared infrastructure creates comparable dependencies in your system, not to assign its harm to an unrelated device.
Cross-check: If this view is not appropriate, record why. The guidance allows an explanation instead of an inapplicable view and does not expect duplicate documentation where another view captures the same risks.
Updateability and patchability architecture document
Check: Show the entire update path, including systems outside the manufacturer's control, authorization, integrity checks, installation, and recovery behavior.
Published FDA scenario: Section V.B.2(c) describes software traveling from an update server to an in-clinic cardiac implant programmer that programs an implanted device. The guidance calls for end-to-end consideration of that path, including intermediary technology the manufacturer does not control.
Document layout: Update stage | sending and receiving asset | authorization/integrity mechanism | installation precondition | interrupted-update behavior | recovery evidence. Record how the actual design handles a failed transfer and how installed versions are identified.
Cross-check: The diagram, secure-update specification, testing, and user update instructions should describe the same mechanism. Do not invent rollback support or a delivery interval.
Security use case architecture document
Check: Describe security-relevant operational and clinical states, not only the network when the device is idle.
Source example: Section V.B.2(d) names power on, standby, and transition states, alongside programming, alarming, delivering therapy, data transfer, and diagnostic reporting. It also notes that shared communication protocols may support combined views, even when the clinical risk assessments differ.
Document layout: Use case and state | actor | permitted action | exchanged information | authentication/authorization | failure behavior | safety-risk reference. Choose use cases established by the actual intended use and architecture.
Cross-check: State transitions and failure handling should connect to the corresponding requirements and tests. A generic login diagram cannot stand in for every clinical function.
Threat model document
Check: Record the method, assets, trust boundaries, threat scenarios, assumptions, and proposed mitigations across the device lifecycle.
Published FDA assumption: Section V.A.1 gives the example of treating hospital networks as hostile and assuming an adversary can alter, drop, and replay packets. That concrete assumption creates questions about message integrity, replay handling, and loss of communication. It does not prove a particular device is vulnerable.
Document layout: Asset/flow | attacker capability and precondition | threat action | affected security objective | potential device effect | proposed control | risk-assessment reference. STRIDE or attack trees can organize analysis; neither supplies a clinical-risk score.
Cross-check: Cover applicable supply-chain, deployment, maintenance, update, and decommissioning paths. Update the model when testing reveals an unmodeled scenario.
Cybersecurity risk assessment document
Check: Evaluate identified scenarios before and after controls, using the manufacturer's declared method and device-specific patient-harm evidence.
Source example: The guidance's simultaneous-monitor-restart scenario distinguishes a multi-patient consequence from a single-device interruption. Its clinical context illustrates why a technical vulnerability rating alone cannot establish patient risk.
Document layout: Threat reference | exploitation evidence and assumptions | device/security impact | safety-hazard link | evaluation method | implemented control | residual assessment | manufacturer decision. Record unavailable evidence as a gap rather than assigning a convenient score.
Cross-check: CVSS describes technical characteristics. It does not provide clinical severity, accept residual risk, or substitute for the manufacturer's safety analysis.
Security requirements and control specification
Check: Describe implemented security behavior precisely enough to evaluate it. Address applicable authentication, authorization, cryptography, logging, recovery, and update controls.
Published FDA control example: Appendix 1 recommends authenticating firmware/software and verifying authentication tags before installation and execution. It also discusses cryptographically signed updates to help prevent downgrade or rollback attacks. These are source recommendations, not claims that your product implements them.
Document layout: Requirement | linked threat | actual control mechanism | protected asset | operating/failure condition | test criterion | evidence reference. For an update control, identify where verification occurs, how verification credentials are managed, and what the device does when verification fails.
Cross-check: “Encrypted” or “secure update” is not a sufficient implementation description. Record actual algorithms, key-management behavior, configuration, and supporting evidence where relevant, without publishing secrets.
SBOM Requirements: Format, Fields, and the Support Report
For applicable cyber devices, Section 524B requires an SBOM covering commercial, open-source, and off-the-shelf software components. The FDA recommends machine-readable presentation in an industry-accepted format. SPDX and CycloneDX are common implementation choices, not exclusive FDA-approved formats.
Baseline elements and component evidence
The February 3, 2026 FDA guidance references the NTIA 2021 baseline attributes: supplier name, component name, component version, other unique identifiers, dependency relationships, the author of the SBOM data, and a timestamp. A newer general SBOM specification does not automatically amend that guidance or Section 524B. Use the FDA SBOM Requirements for Medical Devices to distinguish statutory duties from recommended fields. Use CycloneDX vs. SPDX for Medical Devices for format differences.
Include component support-level and end-of-support information recommended by the FDA. Assess known vulnerabilities for device-specific safety and security impact, with documented disposition and remediation rationale. Supporting information can be linked to SBOM entries without forcing every assessment into the inventory file itself.
Explain how the SBOM was produced
Record the device build, included components, dependency scope, tools, and known limitations. A support report can explain how the inventory was generated and how it will be maintained. This makes the relationship between the submitted software and the component evidence inspectable.
A human-readable PDF or spreadsheet can accompany the machine-readable artifact. It helps readers inspect the inventory but does not replace the recommended machine-readable version. Do not claim that a PDF alone automatically causes a technical-screening hold.
Machine-readable SBOM document
Check: Identify the inventory's software build, component scope, relationships, identifiers, and generation method. Check the chosen format using its tooling.
Actual public format example: The SPDX project's software example1 SBOM describes a small software package named hello. The following excerpt is from that published sample, not a medical-device inventory:
SPDXVersion: SPDX-2.2
DocumentName: hello
PackageName: hello
SPDXID: SPDXRef-Package-hello
Relationship: SPDXRef-DOCUMENT DESCRIBES SPDXRef-Package-hello
Relationship: SPDXRef-hello-binary GENERATED_FROM SPDXRef-hello-src
The complete sample also includes file checksums and license information. Its older format illustrates identifiers and relationships; it is not a complete current medical-device SBOM baseline, a submission template, or evidence of regulatory review.
Cross-check: Replace the sample with an inventory generated from the actual build. Blue Goat uses SPDX as its default delivery format, not an FDA mandate. A component absent from the submitted build should not enter the inventory simply because it appears in a format example.
SBOM support and vulnerability assessment report
Check: Explain inventory generation, tool versions/configuration, scope limits, component support, known vulnerabilities, and maintenance responsibility.
Source example: Section V.A.4 recommends component support-level and end-of-support information alongside vulnerability assessment. A component name in an SBOM does not establish the supplier's support date or whether a vulnerability affects the device.
Document layout: SBOM component reference | supplier support evidence | support status/end-of-support | vulnerability identifier and source | affected-version/reachability assessment | device impact | disposition | evidence/reviewer. Keep unknown supplier dates explicitly unknown pending supplier response.
Cross-check: Reconcile available source/build and binary evidence, and distinguish a scanner match from a reviewed finding. Optional VEX communication needs status-specific evidence and manufacturer review; an SBOM match alone cannot establish “not affected.”
Cybersecurity Test Reports: What Reviewers Examine
Testing evidence should show how the device's security requirements and mitigations were evaluated. Select methods appropriate to the attack surface, which may include penetration testing, vulnerability testing, fuzzing, static analysis, and other security testing. No single method substitutes for all applicable evidence.
For each report, identify the tested configuration, scope, methods, results, findings, and limitations. Connect relevant findings to threats and controls, and document remediation and retest evidence where changes were made. Explain unresolved findings and retain manufacturer-owned residual-risk decisions.
The FDA recommends independent penetration testers. Qualified internal staff independent of development may meet that recommendation; an outside provider is not universally required. Blue Goat provides cybersecurity testing, not general software verification and validation services.
Avoid treating every newly discovered test finding as proof the original threat model was complete. Update the threat model and risk assessment when testing reveals a previously unidentified scenario. The evidence should reflect what the team learned, not force new findings into unrelated threat entries.
Security test plan and requirements test report
Check: Define scope, configuration, methods, test criteria, and exclusions before reporting the actual results. Distinguish planned tests from executed evidence.
Source example: Section V.C recommends evidence for malformed/unexpected inputs, fuzz testing, attack-surface analysis, vulnerability chaining, and static/dynamic analysis. Appendix 1's firmware-authentication recommendation supplies a concrete control topic to evaluate: rejection of software whose authentication verification fails.
Document layout: Requirement/control | test precondition | method and configuration | expected security behavior | observed result | evidence location | finding/deviation. Record actual traffic, logs, or other supporting evidence. Do not mark a control as effective because a procedure exists.
Cross-check: Coverage needs to match applicable interfaces and control behavior. If fuzzing or another method is omitted, explain the scope decision instead of implying it ran.
SAST and software composition analysis report
Check: Identify source-code and component scope, tools, configuration, exclusions, raw results, human review, and confirmed findings.
Published FDA test example: Section V.C specifically names testing for hardcoded, default, easily guessed, or easily compromised credentials. It also recommends software composition analysis of binary executable files. These are concrete analysis topics, not reports that credentials were found in your device.
Document layout: Analysis method | code/build scope | tool/version/settings | raw result | reviewer disposition | confirmed CWE or component vulnerability | risk/remediation link | unresolved-anomaly reference. SAST examines in-scope manufacturer source and modified third-party code; SCA examines components and published vulnerabilities.
Cross-check: Confirm or reject raw results through human review. A scanner export alone does not establish exploitability, remediation, or the absence of security issues.
Penetration test report and retest record
Check: Retain the report's tested configuration, interfaces, exclusions, tester independence and expertise, actual duration, methods, findings, and observations. Preserve an original third-party report where one was used.
Source example: Section V.C explicitly lists those penetration-test report elements. The guidance also recommends recording the manufacturer's assessment of findings, including rationales for deferred remediation. A proposal promising future testing is not that report.
Document layout: Finding reference | affected interface/build | prerequisites | reproducible method and evidence | observed security effect | manufacturer impact assessment | remediation | retest configuration/result | remaining limitation. Populate observations from the executed assessment, never from an invented “passed” example.
Cross-check: The retest must identify what changed and what was reassessed. Qualified internal testers independent of development may meet the independence recommendation; an external provider is not universally required. Blue Goat's Medical Device Penetration Testing covers security testing within the agreed scope.
Unresolved anomaly assessment document
Check: Assess unresolved software anomalies for deliberate exploitation as well as ordinary operation, and record the manufacturer's disposition.
Published FDA scenario: Section V.A.5 describes an anomaly that appears sporadically during normal use but could be triggered continuously by a threat actor. That difference can change the security and patient-safety assessment. It is not evidence of a particular defect or a reason to assign an unsupported risk value.
Document layout: Actual anomaly reference | affected configuration | observed behavior | exploitation analysis/CWE where applicable | safety/security impact | control or rationale | release plan | manufacturer decision. Link deferred remediation to its actual plan and supporting risk rationale.
Cross-check: Keep software-anomaly and cybersecurity records consistent. See Unresolved Anomalies in FDA Cybersecurity for the evidence relationships.
Labeling and Postmarket Documents: What Each Should Contain
Labeling explains the user's security responsibilities; postmarket records explain the manufacturer's ongoing work. They must agree on supported configurations, update behavior, and operational assumptions.
Cybersecurity labeling and user instructions
Check: Tailor security information to the actual users and operating environment. File and reference it according to the applicable labeling prompts.
Published FDA labeling example: Section VI.A lists network ports/interfaces, their incoming or outgoing direction, functionality, and destination endpoints. It also names supporting infrastructure, password requirements, update information, and user responsibilities as topics to consider. These are not instructions to expose secrets or add unneeded ports.
Document layout: Interface or user task | actual configuration | purpose/direction | security responsibility | understandable instructions | limitation/support reference. Use the architecture's interface inventory to build the network table, then check the instructions against implemented and tested behavior.
Cross-check: Do not tell a user to manage a risk they cannot reasonably manage. A promise that all updates install automatically would conflict with evidence describing manual installation.
Postmarket cybersecurity management and disclosure plan
Check: Document monitoring sources, intake, vulnerability assessment, coordinated disclosure, updates/patches, communications, responsibilities, and review triggers.
Source example: Section 524B(b)(1) calls for a plan to monitor, identify, and address postmarket vulnerabilities and exploits, including coordinated vulnerability disclosure and related procedures. Section 524B(b)(2) distinguishes regular-cycle updates for known unacceptable vulnerabilities from out-of-cycle updates for critical vulnerabilities that could cause uncontrolled risks. Those statutory duties do not supply one universal numerical patch deadline.
Document layout: Information source/intake channel | triage owner | affected-version assessment | safety/security decision | update/disclosure path | customer communication | closure evidence. Use a real maintained intake channel and manufacturer-approved processes, not invented contact information or response times.
Cross-check: Include different fielded versions when an update is not universally installed. Link inventory maintenance, labeling, update architecture, and the manufacturer's risk-management process.
Cybersecurity metrics report
Check: Report available observations with definitions, source records, measurement windows, and limitations. Keep planned monitoring separate from measured performance.
Published FDA metric examples: Section V.A.6 recommends tracking the percentage of identified vulnerabilities patched, time from identification to patch availability, and time from patch availability to field implementation to the extent known. It acknowledges that such measures may not be available for a new device without marketed history.
Document layout: Metric definition | period | source records | inclusion/exclusion rules | observed result | field-data limitation | process action. For a patch-percentage measure, identify the numerator and denominator from actual records. No sample percentage or time is supplied here because those values require manufacturer evidence.
Cross-check: “No field history” is not a measured result of zero vulnerabilities or immediate deployment. State unavailable information plainly.
Assembling the Final Package: Traceability and Submission Checks
A traceability matrix should let a reader follow a threat to its control, test evidence, and applicable submission prompt. Use stable identifiers and document-section references. For example, a relevant wireless-interface threat can link to an authentication requirement, an implemented control, and a test evaluating that control.
Traceability matrix and final evidence index
Check: Connect architecture assets, threats, risk analysis, requirements, controls, SBOM components where relevant, tests, anomalies, and residual-risk decisions to their exact record locations.
Source-grounded example chain: The FDA's hostile-network assumption can lead to an analysis of altered or replayed communications, an integrity/authentication requirement, a documented implementation, and an executed test. This describes an evidence relationship, not a claim that a device uses a particular control or passed that test. Appendix 2 specifically recommends links between diagram elements, hazards/controls, and testing.
| Matrix column | What goes in the actual record |
|---|---|
| Asset and flow | Architecture element and section for the assessed interface |
| Threat and risk | Actual scenario identifier and patient-harm analysis reference |
| Requirement and control | Implemented security requirement and design record |
| Component | Relevant SBOM identifier, when applicable |
| Test evidence | Executed method, configuration, result, and report section |
| Remaining issue and decision | Finding/anomaly, remediation or rationale, and manufacturer review |
| Submission location | Applicable eSTAR prompt, attachment title/revision, and page/section |
Cross-check: Empty cells reveal missing evidence or a relationship requiring explanation. They do not become “passed” merely because the attachment exists. Open referenced documents and verify the sections, configurations, and outcomes before upload.
Before uploading, check that:
- The template version and device configuration are recorded.
- Each applicable prompt has a response and supporting attachment reference.
- Architecture, threat model, SBOM, and testing refer to the same configuration.
- Threats and controls link to test evidence, with gaps and limitations identified.
- Known vulnerabilities and unresolved anomalies have documented device-level assessments.
- Metrics distinguish observed results from planned monitoring.
- Labeling and postmarket plans agree on support assumptions and user responsibilities.
- Filenames, document titles, versions, and section references are consistent.
Technical screening and substantive review answer different questions. Complete uploads do not establish evidence adequacy or predict clearance. Keep a record of open gaps and manufacturer decisions rather than presenting an unchecked checklist as a regulatory conclusion.
For package support, see the FDA Premarket Cybersecurity Package. For the eSTAR attachments, see Blue Goat's eSTAR cybersecurity documentation support.
FAQ
What eSTAR 510(k) cybersecurity documentation should we prepare?
Prepare evidence against the applicable template prompts and the device's cybersecurity risks. The evidence generally spans risk management, threat modeling and architecture, SBOM information, controls, testing, labeling, and postmarket plans. Section 524B adds duties for applicable cyber devices. Record each prompt's supporting document and section, and review the connections between artifacts. The number of files is a packaging choice, not a universal FDA requirement.
Does the FDA require exactly 14 or 18 cybersecurity documents?
No universal document count follows from those organizational checklists. The applicable eSTAR prompts, statutory duties, and device-specific evidence determine what to prepare. Several artifacts can share an attachment when references are clear and the relevant prompts are addressed. A checklist can organize preparation, but completing a fixed number of files does not establish evidence adequacy or determine the review outcome.
Is a missing SBOM element automatically a technical-screening hold?
Do not assume a particular omission automatically produces a hold. Technical screening checks completeness, while substantive review evaluates the evidence. Section 524B requires an SBOM for applicable cyber devices, and the FDA recommends machine-readable presentation. Prepare current baseline elements and supporting component information, identify limitations, and consult the applicable template. A human-readable companion can help reviewers inspect the inventory without replacing the recommended machine-readable artifact.
What if a device modification has no cybersecurity impact?
Document an impact analysis before choosing the applicable evidence. Review changes to the architecture, components, connectivity, threats, controls, and testing. A statement that no security features changed does not establish that cybersecurity is unaffected. Use the current applicable template and guidance to identify the modification's documentation needs, and retain the rationale connecting that determination to device-specific records.
How should metrics and unresolved anomalies be documented?
Address the applicable prompts using actual device evidence. Distinguish observed cybersecurity measurements from planned postmarket monitoring, particularly when the device has no fielded history. Assess unresolved software anomalies for security and patient-safety impact, and link relevant anomalies to threats, controls, testing, and residual-risk rationale. Neither a generic bug list nor invented monitoring data provides that connection.
How detailed should the traceability matrix be?
Use stable threat, requirement, control, and test identifiers with document and section references. A reader should be able to locate the evidence supporting a mitigation and understand the tested device configuration. Record findings, remediation, retests, and relevant residual-risk decisions. Update the matrix when testing or design changes reveal new scenarios, rather than forcing every result into a pre-existing threat entry.
Do the attachment groupings represent exact eSTAR 7.1 field names?
No. The evidence groups are a preparation aid. Use the actual prompts and attachment fields in the template for your device and pathway. Several artifacts may share a field, and clearly referenced evidence can serve multiple prompts. Record the attachment title, revision, and section for each applicable response.
Does eSTAR 7.1 check the quality of our cybersecurity evidence?
The template supports preparation and completeness checks. It does not validate a threat model, establish a control's effectiveness, or accept residual risk. Technical screening and substantive review are different. The manufacturer should review the underlying evidence and keep open gaps visible before filing.
Where should cybersecurity labeling go?
Respond to the applicable labeling prompts and cross-reference security information from the cybersecurity evidence where relevant. User instructions should agree with the architecture, tested update behavior, support information, and postmarket plan. Do not assume a fixed section number is valid across every template or pathway.
Should an older submitted eSTAR be replaced with version 7.1?
Follow the FDA's instructions for the existing submission and response type. Do not migrate an acknowledged submission solely because a newer template exists. For a new draft, check the current applicable template and version history. Template changes and changes to the device are separate questions.
How Blue Goat supports eSTAR documentation
Blue Goat reviews device-specific cybersecurity records, identifies gaps, and maps agreed evidence to applicable eSTAR prompts. We connect architecture, threats, controls, security testing, and manufacturer-owned safety references. The manufacturer retains product decisions, clinical harm severity, risk acceptance, and submission responsibilities.
The threat model and Threat Table are delivered together. Security architecture views are scoped and priced separately, even when their evidence is cross-referenced in the same submission.
For scoped support, see eSTAR cybersecurity documentation support and the FDA Premarket Cybersecurity Package. Fees are fixed after scoping. If the FDA raises a cybersecurity question, we answer it at no extra cost until the cybersecurity portion is closed.
Sources and primary references
- FDA final premarket cybersecurity guidance, February 3, 2026. Sections V.A, V.B, V.C, VI, VII, and Appendices 1 and 2.
- FDA eSTAR Program. Applicable templates, versions, and filing instructions.
- FDA Section 524B FAQs.
- CISA SBOM resources.
- SPDX public software example. An older format example, not a medical-device submission.
Related references and tools
- FDA Premarket Cybersecurity Deliverables and eSTAR Map: the separate field-placement reference.
- eSTAR 7.1 Submission Checklist: broader submission preparation.
- eSTAR Software Documents and Cybersecurity: software-section consistency.
- eSTAR Cybersecurity Checklist: an organizational aid, not a regulatory decision.
Sources & references
Primary sources cited in this article. Links open in a new tab.



