Class 2 medical device submission: a practical guide to 510(k), De Novo, and eSTAR

A Class 2 medical device submission sits at the intersection of regulatory strategy, evidentiary discipline, and operational execution. Choosing the wrong submission type, citing an unsuitable predicate, or submitting a structurally inconsistent eSTAR package can add months to your timeline before a single substantive question is answered.

This guide is written for regulatory affairs and regulatory operations professionals. It helps you make defensible pathway choices, build an eSTAR-ready dossier, and sequence post-clearance obligations — all under current FDA expectations.

Overview

This section ties your submission choice to the regulatory classification that defines required evidence. It helps you decide whether a 510(k), De Novo, or an exemption applies before drafting begins.

Class II devices are moderate-risk devices for which general controls alone are insufficient. Special controls — performance standards, postmarket surveillance requirements, patient registries, and device-specific guidances — provide reasonable assurance of safety and effectiveness when combined with general controls. As FDA's classification guidance states, if a device is classified as Class I or II and is not exempt, a 510(k) is generally required before marketing. Demonstrating "substantial equivalence" to a legally marketed predicate device through a Premarket Notification (510(k)) is the standard path to clearance for most Class II products.

Special controls are codified at the product-code level and substantially define what evidence your submission must contain. Identifying your correct regulation number and product code before drafting substantive sections is therefore essential — it determines the testing standards, labeling language, and clinical data expectations you must address.

Where no suitable predicate exists, De Novo classification allows a sponsor to establish new special controls and create a predicate for future devices. Some Class II devices remain exempt from 510(k) requirements, subject to explicit limitations on indications and specifications.

Pathway decision framework for Class II devices

Choose the correct submission pathway early. This decision determines evidence scope, review dynamics, and downstream obligations, and reversing course mid-dossier is costly.

FDA recognizes Traditional, Special, and Abbreviated 510(k)s, with De Novo as a parallel pathway for novel moderate-risk devices. Each pathway has distinct eligibility criteria and evidence expectations that should be confirmed against current FDA guidance before committing to a strategy.

The Traditional 510(k) is the default when substantial equivalence must be demonstrated through a full comparison and performance data. The Special 510(k) applies when intended use is unchanged and design controls suffice for certain modifications to your own cleared device. The Abbreviated 510(k) leverages Declarations of Conformity to FDA-recognized standards in place of detailed test data. De Novo is appropriate for novel moderate-risk devices lacking an adequate predicate; it requires proposing special controls and a classification regulation rather than a substantial equivalence comparison.

Worked example — pathway selection in practice: A regulatory affairs lead at a mid-stage company prepares a submission for a wireless pulse oximeter with Bluetooth connectivity and an embedded algorithm for trend detection. Cleared predicates exist within the same product code family and the intended use — non-invasive monitoring of oxygen saturation — is the same.

However, the connectivity and embedded algorithm introduce software level-of-concern and cybersecurity implications that change the risk profile relative to those predicates. Because the technology raises new safety and effectiveness questions, the team selects a Traditional 510(k) rather than an Abbreviated pathway. They plan performance testing per applicable IEC standards and FDA software and cybersecurity guidance, and schedule a Pre-Submission meeting to align on algorithm classification and evidence expectations before drafting the eSTAR package. The Pre-Sub investment front-loads alignment cost and reduces the probability of AI requests later.

Traditional, Special, and Abbreviated 510(k): eligibility at a glance

The Traditional 510(k) requires a full substantial equivalence comparison covering intended use, technological characteristics, and performance data, along with all relevant FDA forms, device description, labeling, and testing summaries. Use it when new technological characteristics or modified intended uses are present.

The Special 510(k) is limited to modifications to your own legally marketed device where the modification does not affect intended use, and design controls serve as the primary evidence. Modifications that raise new questions of safety or effectiveness are ineligible and should use a Traditional 510(k) instead — see the FDA Special 510(k) Program guidance for the current eligibility criteria and documentation expectations.

The Abbreviated 510(k) relies on conformity to one or more FDA-recognized consensus standards, with Declarations of Conformity replacing detailed test data. Eligibility requires that the cited standards directly address the device's safety and effectiveness questions.

Pathway selection signals — at a glance:

• Traditional 510(k): new device or modification with new technological characteristics; full substantial equivalence comparison and performance data required.
• Special 510(k): modification to your own cleared device; no intended use change; design controls are primary evidence; no new safety or effectiveness questions raised.
• Abbreviated 510(k): device characterization fully addressed by FDA-recognized standards; Declarations of Conformity available and applicable.
• De Novo: novel moderate-risk device with no adequate predicate; sponsor willing to propose special controls and establish a new classification regulation.

When De Novo is preferable to a weak predicate

Use De Novo when forcing equivalence to an outdated or poorly matched predicate would introduce legal and strategic risk. FDA reviewers evaluate whether intended use matches and whether technological differences raise new questions of safety or effectiveness that performance data can resolve — a fragile equivalence argument may survive review or it may not, and the uncertainty itself carries cost.

Predicates cleared decades ago or under obsolete standards often produce weak substantial equivalence arguments. De Novo allows a sponsor to establish device-appropriate special controls that reflect the current risk profile and, importantly, creates a clean predicate for future products in the same category. The FDA De Novo classification guidance describes the required elements, including a proposed regulation text and draft special controls.

The trade-off is longer review timelines and the obligation to draft a proposed classification regulation. For platform-building sponsors, De Novo often justifies the upfront investment. For one-off products with close predicates and well-characterized evidence, a Traditional 510(k) generally remains preferable.

How to identify your classification regulation, product code, and special controls

Start evidence planning by confirming the regulation number and product code. These determine the applicable special controls and device-specific expectations before a single testing protocol is written.

Use FDA's Product Classification Database to search by device type, intended use, or medical specialty. Each record lists the regulation number under 21 CFR Parts 862–892, the product code, device class, and required submission type. Retrieve the classification regulation text from the Code of Federal Regulations to identify the special controls, which are the mandatory starting points for your evidence plan.

Device-specific guidance documents published by FDA's CDRH commonly supplement the regulation text with testing thresholds, acceptance criteria, labeling requirements, and clinical data expectations. Map each special control to a specific eSTAR section. If a special control requires testing to a particular standard, include that standard in your evidence section and provide either a Declaration of Conformity or a testing summary addressing each requirement. Ensure labeling language required by special controls appears verbatim in your draft IFU — deviations from specified language are a frequent RTA trigger.

Cross-reference FDA's Recognized Consensus Standards Database to identify applicable recognized standards and prepare Declarations of Conformity where available to streamline evidence while maintaining reviewer confidence.

Preparing a Class II 510(k) in eSTAR

Frame eSTAR workstreams early. The template prescribes submission structure and constrains narrative flexibility — assembling eSTAR-consistent content before upload reduces rework and RTA risk substantially.

eSTAR is FDA's mandatory electronic submission template for 510(k)s. It organizes the submission into defined fields with embedded instructions and cross-reference logic. Key administrative forms — FDA Form 3514 and FDA Form 3881 — are completed inside the template and cannot be treated as standalone attachments. See FDA's eSTAR resources page for the current template and instructions.

The template aligns with FDA's review checklist: device description, substantial equivalence comparison, labeling, performance testing, software documentation, and biocompatibility. Cross-sectional consistency is essential. Inconsistencies between Form 3881 indications and the substantial equivalence narrative, or between the device description and the comparison table, are among the most common causes of RTA decisions. For complex submissions involving software, cybersecurity, and human factors elements, assigning shared ownership of eSTAR sections early — with version-controlled tracking of who owns each section and what has changed — prevents the kind of cross-section drift that generates last-minute remediation work.

eSTAR completion checklist — core sections for a Class II 510(k):

• FDA Form 3514 (cover sheet): complete and consistent with body sections
• FDA Form 3881 (Indications for Use): identical language to device description and substantial equivalence comparison
• 510(k) Summary or Statement elected; Summary consistent with comparison section
• Device description: components, materials, software versions, accessories
• Predicate identification: cleared 510(k) number, device name, cleared indications, selection rationale
• Substantial equivalence comparison: intended use and technological characteristics, addressing each difference with rationale or data
• Performance testing: bench testing summaries with acceptance criteria; clinical data summary if applicable
• Biocompatibility: ISO 10993 assessment or testing summary aligned to contact duration
• Software documentation: level of concern, description, hazard analysis, V&V summary, cybersecurity documentation as applicable
• Human factors: usability engineering summary or rationale for no summative study
• Labeling: draft label and IFU addressing special controls
• Declarations of Conformity: standard number, revision, and scope of conformance
• Sterility and packaging documentation: if applicable per ISO sterilization and packaging standards

Common eSTAR RTA triggers and how to avoid them

Prevent RTA by resolving common acceptance-level deficiencies before submission. Many RTA decisions stem from problems that a structured pre-submission review would catch.

Frequent issues include inconsistent indications for use across Form 3881, device description, and labeling; missing or incomplete labeling; and absent or incomplete Declarations of Conformity. Omitted or misclassified software documentation — including missing level-of-concern determinations — is a recurring cause, as are incomplete biocompatibility assessments with mismatched contact duration classifications and cross-reference inconsistencies in the substantial equivalence comparison. For connected devices, missing cybersecurity documentation is an increasingly common trigger.

Address these by performing a final cross-reference and consistency pass before upload: verify all DoCs state the standard version and scope, confirm software and cybersecurity artifacts meet current FDA cybersecurity premarket guidance expectations, and check that draft labeling contains all special control language verbatim.

Evidence and testing matrix for common Class II device archetypes

Align testing plans to the device archetype and to recognized standards so each identified risk is addressed by a test or a documented rationale for why testing is unnecessary.

Patient-contacting non-electrical devices require biocompatibility evaluation per ISO 10993, guided by FDA's Use of ISO 10993-1 guidance, and mechanical performance testing matched to the risk analysis. Labeling must cover contact duration, cleaning instructions, and reprocessing instructions where applicable.

Electro-medical devices need electrical safety and EMC testing per IEC 60601-1 and applicable collateral and particular standards, software documentation if software controls device functions, and human factors documentation where use-related risks exist.

Sterile devices require sterilization validation per the applicable ISO 11135, ISO 11137, or ISO 17665 standard (depending on sterilization modality), packaging validation per ISO 11607, shelf-life stability data, and sterility-related labeling.

Software-enabled devices and Software as a Medical Device require a level-of-concern determination, hazard analysis, and V&V summaries; IEC 62304 lifecycle documentation for moderate- or high-concern software; cybersecurity artifacts including an SBOM and a threat model consistent with FDA premarket cybersecurity guidance; and human factors engineering documentation.

Across all archetypes, ISO 14971 risk management artifacts should map explicitly to test protocols, labeling controls, and design features. Reviewers look for integration — a risk file that references specific test protocols and labeling mitigations is more defensible than one maintained in isolation from the submission.

Predicate selection strategy and building a substantial equivalence comparison

Predicate selection is a strategic decision that underpins the entire substantial equivalence argument. Document your search and selection logic as thoroughly as you document the comparison itself.

Search FDA's 510(k) database for cleared devices within your product code that share intended use. Evaluate candidates against intended use similarity, technological similarity, recency of clearance, and the quality of publicly available clearance information. A predicate cleared recently with a thorough publicly available decision summary is generally more defensible than an older predicate with limited documentation.

Construct a defensible comparison table organized around intended use and technological characteristics. For intended use, compare clinical purpose, patient population, indication, and use environment side by side. For technological characteristics, compare design, materials, energy source, operating principles, and software architecture item by item.

For each technological difference, either argue that the difference does not raise new questions of safety or effectiveness with supporting rationale, or cross-reference performance data demonstrating equivalent safety and effectiveness. An assertion without supporting data or rationale is incomplete and will invite an AI request.

Predicate comparison template — structural outline:

• Header: subject and predicate device names and 510(k) numbers
• Intended use comparison: side-by-side statements and conclusion with rationale
• Technological characteristics comparison: item-by-item (design, materials, operating principle, software, sterility, connectivity) with either rationale that the difference does not raise new safety or effectiveness questions, or cross-reference to performance data
• Summary conclusion: statement of substantial equivalence citing predicate 510(k) number

Avoid split predicates unless necessary and rigorously justified. When clinical data are included, explicitly bridge their scope to the substantial equivalence comparison and special controls so reviewers do not need to infer the connection.

Timelines, review stages, and fees under MDUFA V

Plan internal readiness checkpoints to protect the FDA review clock and meet MDUFA V performance commitments on both sides of the submission.

Under MDUFA V performance goals, FDA aims to complete acceptance review within approximately 15 calendar days and to reach a substantive review decision on Traditional 510(k)s within 90 FDA days. An RTA resets the acceptance clock; an Additional Information (AI) request pauses the substantive-review clock until FDA receives your response. Both outcomes are avoidable with pre-submission discipline.

User fees apply to most 510(k)s and are published annually on FDA's Medical Device User Fee page. Small businesses may qualify for reduced fees under FDA's Small Business Determination program — applying before paying is time-sensitive, and the determination must typically be in place before the submission is filed.

Internal readiness checkpoints before submission:

• Classification confirmed and product code verified
• Predicate or predicates identified and comparison table complete
• eSTAR populated and cross-references verified
• Draft labeling reviewed against special controls
• Declarations of Conformity completed and signed
• Software documentation package complete
• Biocompatibility assessment aligned with device description
• User fee payment confirmed or Small Business Determination obtained
• QMS documentation consistent with submission representations

Managing the FDA clock: acceptance, interactive review, and AI pauses

Eliminate RTA risk through pre-submission quality checks, because an acceptance-level deficiency resets the clock and typically adds weeks. After acceptance, be prepared for AI requests that pause the FDA review clock until your response is received — FDA's performance goal clock stops running while the response is pending.

Interactive review can reduce AI cycles by enabling informal communication with reviewers, but responses must maintain evidentiary rigor to avoid formal AI issuance. Plan internal response timelines of approximately two to three weeks for AI and interactive review responses and build those windows into your project schedule before submission. Teams that treat AI responses as improvised reactions rather than planned deliverables consistently lose more clock time than those who pre-assign ownership and draft response templates.

Q-Submissions and 513(g): when and how to use them

Use Pre-Subs and 513(g) requests to de-risk classification and strategy questions before drafting a full dossier. Both tools are underused relative to the rework they can prevent.

A Pre-Submission (Pre-Sub) is valuable when technology is novel, predicate selection may be contested, planned testing may be insufficient, or software and cybersecurity approaches need alignment with current expectations. Frame Pre-Sub questions narrowly and specifically — vague questions produce vague guidance. FDA's Q-Submission Program guidance describes meeting request formats and FDA's response commitments.

A 513(g) Request for Information asks FDA to identify classification and regulatory requirements for a described device. It is appropriate when you are uncertain about Class I/II/III status or want a formal written classification position before investing in testing and dossier assembly.

The practical sequencing rule is straightforward: use 513(g) to confirm classification, then use a Pre-Sub to align on submission approach once classification is known.

Third-Party 510(k) Review: eligibility and tradeoffs

Consider Third-Party Review when your device is eligible and the evidentiary issues are straightforward. It can shorten review timelines by routing the submission through an FDA-accredited reviewer rather than directly through CDRH staff review.

Third-Party Review is limited to eligible device types and excludes Class III devices and those removed from the program. Confirm eligibility against FDA's current list before committing, as the eligible device set changes periodically.

The submitter pays the accredited reviewer directly and still pays an FDA user fee, which may be reduced for eligible submissions. FDA retains authority to disagree with the accredited reviewer's conclusions, which can require additional FDA review and extend the overall timeline. Third-Party Review is strategically attractive for straightforward devices with clean predicates and well-characterized evidence — the benefit narrows for complex submissions where substantive reviewer judgment is likely to matter.

Software and cybersecurity expectations for Class II devices

Treat software and cybersecurity documentation as core evidentiary elements for connected and software-enabled devices. These sections commonly prompt AI requests when incomplete or when they do not reflect current FDA expectations.

FDA's level-of-concern framework — Major, Moderate, or Minor — determines documentation depth. Major-concern software requires comprehensive software documentation including full V&V records. Moderate concern requires a defined subset. Minor concern requires minimal documentation. The level-of-concern determination itself must be documented and justified.

For connected devices, FDA's cybersecurity premarket guidance and Section 524B of the FD&C Act establish cybersecurity plan requirements covering monitoring and mitigation of vulnerabilities throughout the device lifecycle. Practical expectations include a Software Bill of Materials (SBOM), a threat model, a security risk assessment using NIST or MITRE frameworks, and a postmarket patch and update plan. Submitting without these artifacts for a connected device is a predictable AI trigger.

Devices incorporating AI and ML algorithms are reviewed under a framework that currently expects algorithms to be locked at clearance for standard 510(k) pathways. Continuous-learning or adaptive models raise additional questions best resolved through a Pre-Sub meeting before committing to a submission strategy.

Human factors and labeling: integrating user risk into the submission

Integrate human factors engineering evidence with labeling and risk management so use-related risks and their controls are explicit and traceable throughout the submission — reviewers should not need to infer the connections.

HFE documentation should include a use-related risk analysis mapping critical tasks to hazards and risk controls, a description of the user interface and the design rationale behind it, formative evaluation results where applicable, and summative usability testing results with participant demographics and critical task outcomes. FDA's human factors guidance describes the documentation expectations in detail.

Labeling — the device label and IFU — must directly address each critical task and each use-related risk control identified in the HFE analysis. When special controls specify labeling language, include that language verbatim in the IFU. Labeling must also meet general controls under 21 CFR Part 801; unclear or insufficient labeling is a recurring substantial equivalence concern and a frequent post-acceptance AI subject.

Changes to a cleared Class II device: when a new 510(k) is required

Assess every modification under FDA's 2017 guidance on when to submit a 510(k) for a change to an existing device. Document the evaluation and retain it in the Design History File regardless of conclusion.

Changes likely to require a new 510(k) include intended use changes affecting patient population or indication, new technological principles involving different operating mechanisms or core algorithms, material substitutions with new biocompatibility implications, software changes affecting intended use or safety, and sterilization or packaging changes requiring new validation.

Changes limited to labeling corrections, cosmetic changes to non-patient-contacting components, or manufacturing process changes fully validated through design controls typically do not require a new 510(k). The critical point is that the rationale for either conclusion must be documented contemporaneously and retained — undocumented decisions made informally are the most common vulnerability in post-market audit scenarios.

After clearance: UDI/GUDID, registration/listing, and launch sequencing

Sequence post-clearance obligations to avoid launch delays and compliance gaps. Clearance does not alone authorize commercial distribution, and the remaining steps carry their own deadlines.

Assign a UDI and submit the Device Identifier and Production Identifier to FDA's GUDID before commercial distribution. Ensure the UDI appears on device labeling and packaging in both human-readable and AIDC formats as required by FDA's UDI rule.

Register the establishment and list the cleared device in FDA's registration and listing system prior to distribution. Registrations are renewed annually by December 31 each year. Update listings for discontinued devices or changes affecting listed information.

Confirm postmarket obligations and QMS readiness — including MDR reporting procedures under 21 CFR Part 803, recall and correction procedures under 21 CFR Part 806, any Section 522 postmarket surveillance orders attached to the clearance order, and alignment with 21 CFR Part 820 for commercial production.

Post-clearance sequencing summary:

1. Review 510(k) clearance order and any clearance conditions attached.

2. Assign UDI Device Identifier through an FDA-accredited issuing agency.

3. Submit device data to GUDID before commercial distribution.

4. Register establishment and list device with FDA before distribution.

5. Confirm postmarket surveillance, MDR, and recall procedures are documented and resourced.

6. Confirm QMS readiness for commercial production.

On-page utilities

These checklists provide a decision frame you can apply at each major milestone — pathway selection, dossier assembly, and pre-submission — to reach defensible conclusions and avoid acceptance-level deficiencies.

Pathway decision matrix checklist

• Confirm classification regulation number and product code via FDA's Product Classification Database.
• Confirm whether the device is 510(k)-exempt and verify any applicable limitations.
• If not exempt, determine whether a suitable predicate exists with substantially equivalent intended use and manageable technological differences, then select Traditional, Special, Abbreviated 510(k), or De Novo accordingly.
• If De Novo is under consideration, confirm willingness to propose special controls and establish a new classification regulation.
• Consider a Pre-Sub if pathway choice, predicate selection, or software and cybersecurity expectations are uncertain.

eSTAR completion checklist

• FDA Form 3514 completed and consistent with submission content.
• FDA Form 3881 Indications for Use language identical throughout submission.
• 510(k) Summary or Statement elected and consistent with substantial equivalence narrative.
• Device description complete with component list, materials, software versions, and accessories.
• Predicate identified with cleared 510(k) number and comparison complete.
• Performance testing summaries present with acceptance criteria and results.
• Declarations of Conformity present for each cited FDA-recognized standard, with standard version and scope stated.
• Biocompatibility evaluation present and consistent with contact classification.
• Software documentation and cybersecurity artifacts present for connected devices.
• Human factors documentation or written rationale for no summative study present.
• Draft labeling and IFU including verbatim special control language.
• Cross-references verified to resolve correctly within the eSTAR template.

Predicate comparison template outline

• Subject device: name, applicant, proposed indication for use.
• Predicate device: name, applicant, cleared 510(k) number, cleared indication.
• Intended use comparison: side-by-side text and rationale.
• Technological characteristics comparison: design, materials, operating principle, software, connectivity, sterilization; for each difference, provide rationale or cross-reference to performance data.
• Conclusion: statement of substantial equivalence citing predicate 510(k) number.

Timeline readiness checkpoints

• Predicate confirmed and comparison table draft complete: minimum eight weeks before target submission.
• eSTAR populated (first complete draft): six weeks before target submission.
• Internal review cycle complete: four weeks before target submission.
• All testing completed and summarized: four weeks before target submission.
• Labeling finalized and cross-referenced: three weeks before target submission.
• Declarations of Conformity signed: two weeks before target submission.
• User fee payment confirmed or Small Business Determination obtained: two weeks before target submission.
• Final eSTAR cross-reference and consistency check: one week before target submission.
• AI response plan documented and resourced: within two weeks of any AI or interactive review request post-submission.

---

The decision frame running through this guide reduces to three sequenced choices: confirm classification and special controls before drafting; select the pathway that matches your predicate situation and risk profile; and build the eSTAR package with cross-section consistency from the first draft rather than the last. Teams that run a structured pre-submission quality check against the checklist above — covering indications consistency, Declarations of Conformity completeness, software and cybersecurity documentation, and labeling — before uploading to eSTAR consistently face fewer acceptance-level issues. If your submission involves eCTD delivery, software-heavy devices, or multi-reviewer dossier assembly, a controlled workspace with shared version state and automated structural checks — such as Assyro's eCTD submission platform, which validates structural, lifecycle, and metadata conformance continuously as the dossier is assembled — can reduce the late-stage remediation that erodes submission timelines.