FDA De Novo classification is the US pathway for genuinely novel medical devices that carry low-to-moderate risk but have no valid predicate on the US market, which means the standard 510(k) substantial-equivalence route is not available. A De Novo request asks the FDA to create a new classification for the device type — typically Class I or Class II with appropriate special controls — and to clear the specific device at the same time. For EU startups with innovative technology that does not map cleanly to anything already cleared in the United States, De Novo is often the right FDA pathway, and understanding when it fits is a strategic question that belongs at the start of a US plan, not after a failed 510(k).

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

TL;DR

  • De Novo is the FDA pathway for novel devices of low-to-moderate risk that have no valid predicate and therefore cannot use the 510(k) route, but do not warrant the burden of PMA.
  • A successful De Novo request does two things at once: it clears the specific device, and it creates a new device classification that future 510(k) submissions from other manufacturers can reference as a predicate.
  • The analytical core of a De Novo is a risk-and-controls argument built from scratch, not a predicate comparison. The FDA must be convinced that the identified risks can be mitigated through general controls, or general controls plus special controls.
  • For EU startups, De Novo is often the right pathway for innovative software as a medical device, novel digital health tools, and new diagnostic approaches that do not fit an existing US classification.
  • A misread of the pathway question at the start of a US programme is one of the most expensive mistakes first-time EU founders make. A single structured meeting with a US regulatory specialist resolves the question reliably.

Why this pathway matters for EU founders

Most EU founders who start thinking seriously about the United States land on the 510(k) first, because it is the pathway they have heard about. They learn that the 510(k) rests on substantial equivalence to a legally marketed predicate, and then they hit a wall: their device is genuinely new. There is no cleared device on the US market that has the same intended use and comparable technological characteristics. The predicate does not exist.

At that point, one of two things happens. Either the founder concludes the US market is closed to them and gives up, or they try to force a 510(k) against a weak predicate and run into trouble during review. Both outcomes are unnecessary. The De Novo pathway exists precisely for this situation: a novel device of low-to-moderate risk, no predicate, and a risk profile that does not justify the full burden of PMA.

De Novo is less famous than 510(k) and less feared than PMA, which means many founders only discover it after they have already wasted time on the wrong route. This post is the orientation so that does not happen to you. As with everything on the FDA side in this blog, it is written at the general framing level. For the specific De Novo strategy for your device, the specific risks to identify, the specific special controls to propose, and the specific FDA review division that will handle your request, work with a US regulatory specialist who practices inside the FDA system daily.

What De Novo actually is

De Novo is a formal FDA request under which a manufacturer asks the agency to classify a new device type into Class I or Class II and, in the same action, to clear the specific device the request is built around. The name captures the core idea: the classification is being created from scratch, not inherited from a predicate.

A successful De Novo request produces two outcomes in a single FDA action. The first outcome is that the specific device receives marketing authorisation in the United States. The second outcome is that a new device classification enters the FDA's classification database, with a new product code, a risk class (Class I or Class II), and — for Class II outcomes — a set of special controls that describe the risk mitigations the FDA considers necessary for devices of that type. Once the classification exists, future manufacturers with a similar device can use the De Novo device as a predicate for a 510(k).

This is a genuinely different structure from the 510(k). A 510(k) asks "is this new device as safe and effective as an existing one?" A De Novo asks "what are the risks of this new device type, what controls address those risks, and is the specific device in front of you sufficient to justify creating this classification?" The FDA is not comparing against a predicate. It is reasoning about risk from first principles and deciding how the new device type should be regulated going forward.

For the higher-level framing of where De Novo sits among the FDA pathways, see FDA regulation of medical devices: a primer for EU startups. For the 510(k) comparison, see The 510(k) process explained: how to get FDA clearance for your device.

When De Novo fits a novel device

De Novo fits when four conditions line up. Miss any of them and the pathway is probably wrong.

The first condition is genuine novelty. The device must not fit any existing classification in the FDA's classification database. This is not a casual "we think we are new" judgement — it is a formal search across product codes and cleared devices in the relevant space. A device that looks novel from inside the founding team often turns out to have a plausible predicate when a US regulatory specialist runs a proper search. The novelty has to survive that search.

The second condition is low-to-moderate risk. The device's risk profile must plausibly sit in the Class I or Class II range under US logic. A device whose risks clearly push into Class III territory is not a De Novo candidate — it is a PMA candidate. The risk-class question on the US side is answered by the FDA's risk logic, not by the device's MDR class. A device that is Class IIa under the MDR might be Class II or Class III under the FDA depending on how the agency reasons about the specific risks.

The third condition is a workable risk-and-controls argument. The applicant must be able to identify the risks of the new device type and propose controls — either general controls alone, or general controls plus special controls — that address those risks. The quality of this argument is the difference between a De Novo that moves through review and one that stalls. If the risks are real but the controls are speculative, the agency will push back.

The fourth condition is commitment to the pathway. De Novo is more involved than a typical 510(k) because the agency is doing more than comparing to a predicate — it is reasoning about a new device type and, in effect, writing the regulatory framework for that type going forward. That extra analytical burden shows up in the preparation work, in the interaction with the review division, and in the willingness of the applicant to invest in a structured pre-submission conversation with the FDA before filing.

When these conditions align, De Novo is often the cleanest route into the US market for a novel device. When they do not, another pathway is probably the answer.

De Novo compared with 510(k) and PMA

The three main premarket pathways on the FDA side form a rough spectrum of risk and evidentiary burden, and placing De Novo inside that spectrum is the clearest way to see what it is for.

The 510(k) is the comparison pathway. It rests on substantial equivalence to a legally marketed predicate, and the analytical work is built around that comparison. It is the right pathway when a valid predicate exists and the device sits in the Class II risk range. It is the wrong pathway when no predicate exists or when the device is genuinely new.

PMA is the approval pathway for high-risk devices. It demands substantial clinical evidence, typically from one or more well-controlled clinical investigations, and it applies to most Class III devices. It is the most stringent and most expensive of the three pathways, and the timelines are longer. It is the right pathway for high-risk novel devices and the wrong pathway for a device whose risks do not justify that level of scrutiny.

De Novo sits between the two. It is used when no predicate exists (ruling out 510(k)) but the risk profile does not rise to Class III (ruling out PMA). It is not a "510(k) with extra steps" and it is not a "PMA with fewer steps." It is a distinct pathway with its own analytical logic: classify the device type by reasoning about risk and controls, and clear the specific device as part of the same action.

The choice between the three is not a judgement call the founder makes in isolation. It is a structured analysis that a US regulatory specialist runs against the FDA's current classification database, product codes, and review division practice. Getting the choice wrong at the start is one of the most expensive mistakes in US market access, and it is almost always preventable with a single structured meeting early in the project.

Clinical evidence expectations

A De Novo is not a "light" pathway in terms of evidence. The absence of a predicate means the FDA has less to lean on when reasoning about safety and effectiveness, and the agency typically wants to see enough evidence to be comfortable with the risk-and-controls argument the applicant is making.

The specific evidence expectations depend on the device. For many novel low-to-moderate risk devices, bench testing, software verification and validation where applicable, biocompatibility where relevant, and electrical safety and electromagnetic compatibility for electrical devices will form the core of the performance data package. For devices where the novelty touches directly on clinical outcomes — novel diagnostic algorithms, new digital therapeutics, new physiological measurement approaches — the FDA often expects clinical data that supports the specific performance claims.

For EU founders, the key planning point is that clinical data generated for MDR can often feed into a De Novo package, but it rarely transfers without adjustment. The FDA has its own views on trial design, endpoints, statistical analysis, and reference populations, and a clinical investigation designed only with MDR Annex XIV in mind can miss FDA-relevant elements. The fix is to design the clinical strategy with both regulators in mind from the beginning, and to use FDA pre-submission meetings to confirm the evidence expectations before committing to a specific study design.

The risk-and-controls argument itself is the analytical core of the De Novo and belongs at the centre of the evidence plan. The applicant identifies the risks of the new device type, proposes controls, and builds the evidence that supports those controls. Bench data, software V&V, usability data, clinical data, and post-market commitments all feed into this argument. A De Novo submission that buries the risk-and-controls logic under undifferentiated data is harder to review than one that foregrounds the argument and uses the data to support it.

Timeline at general framing

There is no honest single number for De Novo timelines, and anyone who gives you one is speculating. The actual elapsed time depends on the device, the review division, the quality of the submission, whether the FDA issues additional information requests, the use and depth of pre-submission meetings, and the current state of the agency's review performance in the relevant product area.

What is predictable is the shape. A De Novo typically takes longer than a comparable 510(k) because the agency is doing more analytical work: reasoning about a new device type, deciding on a classification, drafting special controls where applicable, and reviewing the specific device. The preparation phase is also typically longer because the risk-and-controls argument needs more development than a predicate comparison.

The practical planning rule: treat a De Novo as a multi-quarter programme from "we have locked the intended use" to "we have the classification and the clearance." Pre-submission meetings with the FDA, used well, can significantly de-risk the programme by surfacing the agency's concerns before the formal submission is filed. For current performance metrics, current user fees, and the current state of De Novo review in your device category, consult the FDA directly and rely on a US regulatory specialist for the interpretation.

For the broader pathway comparison and cost framing, see The 510(k) process explained and The PMA pathway for EU startups.

When EU startups should consider De Novo

De Novo is worth considering when the product is genuinely new to the US market and the United States is a real, budgeted part of the business plan. In Tibor's experience with EU founders, four situations come up repeatedly:

The first is software as a medical device that does something no existing US-cleared SaMD does. New decision-support logic, new algorithmic approaches, new data sources — these frequently fail the predicate search because nothing comparable has been cleared.

The second is novel diagnostic devices that use a new measurement principle or target a new clinical question. The underlying hardware might be conventional, but the intended use and the clinical claim are new enough that a predicate search comes up empty.

The third is new digital health or digital therapeutic tools where the mechanism of action differs meaningfully from anything cleared in the relevant space. "Meaningfully" is doing a lot of work in that sentence, and the judgement belongs to a US regulatory specialist, not to the founding team.

The fourth is innovative combinations — devices that blend hardware, software, and clinical logic in a way that does not map onto any single existing classification. These often look like De Novo candidates once a proper pathway analysis is run.

In all four cases, the same pattern holds. The pathway question should be answered early, with specialist input, before significant development decisions are locked in. The cost of getting the pathway right at the start is small. The cost of getting it wrong and discovering the mistake mid-project is large.

The Subtract to Ship angle

The Subtract to Ship framework — strip every activity that does not trace to a specific regulatory obligation — applies to De Novo work the same way it applies to 510(k) and MDR work. The discipline is to remove speculative effort and keep required effort.

On the De Novo side, the most common waste to subtract is speculative submission drafting before the pathway choice has been validated, evidence generation that does not directly support the risk-and-controls argument, and QMS theatre that is disguised as quality work. The underlying effort — the risk analysis, the controls argument, the performance data, the QMS alignment to 21 CFR Part 820, the submission drafting — still has to happen. Subtraction cuts the regulatory theatre, not the regulatory work. See The Subtract to Ship framework for MDR for the foundational methodology.

For founders running a dual-market EU plus US programme, the framework extends naturally. One intended purpose drafted with both regulators in mind, one QMS aligned to both ISO 13485 and Part 820, one clinical evidence plan designed to serve both MDR Annex XIV and FDA expectations where possible, two submissions prepared in parallel. Serial planning — CE first, then De Novo as an afterthought — adds years to the global timeline and usually costs more than the parallel work.

Reality Check — Is De Novo actually your pathway?

  1. Have you run a proper US predicate search with a regulatory specialist, or are you assuming no predicate exists because you have not found one yet?
  2. Is the risk profile of your device plausibly in the Class I or Class II range under US logic, confirmed by a specialist rather than inferred from your MDR class?
  3. Can you articulate the risks of your device type and the controls that address them, in the form of an argument a reviewer could follow?
  4. Is the intended use locked and consistent with the intended purpose under MDR Article 2(12)?
  5. Have you considered an FDA pre-submission meeting to surface the agency's expectations before filing?
  6. Is your clinical evidence plan designed to serve both MDR Annex XIV and US evidence expectations, or only one?
  7. Is the US part of your plan a real, budgeted programme with a target launch window, or a line item in a pitch deck?
  8. Do you have a US regulatory specialist identified and scoped, or are you still planning to work this out with your EU consultant?

If you cannot answer more than four of these eight clearly, De Novo is a hypothesis, not a plan. That is fine if you own it and fix it early. It is dangerous if investors or the board are hearing a different story.

Frequently Asked Questions

What is the difference between De Novo and 510(k)? A 510(k) rests on substantial equivalence to a legally marketed predicate. A De Novo does not require a predicate — instead, it asks the FDA to create a new classification for the device type and clear the specific device at the same time. You use 510(k) when a valid predicate exists. You use De Novo when the device is genuinely novel, sits in the low-to-moderate risk range, and no predicate is available.

Is De Novo faster or slower than 510(k)? Typically slower, because the FDA is doing more analytical work: reasoning about a new device type, deciding on a classification, and drafting special controls where applicable. The exact timeline depends on the device, the review division, the quality of the submission, and whether additional information requests are issued. Plan in quarters, not weeks.

Can I go from a failed 510(k) into a De Novo? In principle yes — a Not Substantially Equivalent finding on a 510(k) can open the door to a De Novo request — but this is a costly detour compared with picking the right pathway at the start. A single structured meeting with a US regulatory specialist early in the project is almost always cheaper than discovering the pathway was wrong after the fact.

Does an ISO 13485 QMS satisfy the FDA for a De Novo? It is a strong base but not automatically sufficient. The FDA's Quality System Regulation at 21 CFR Part 820 has specific requirements and inspection expectations, and EU manufacturers should plan for a gap analysis and close any Part 820-specific gaps. The current status of the FDA's QMSR harmonisation initiative belongs on the FDA's own guidance pages, not in a primer.

Can I reuse my MDR clinical evidence for a De Novo? Sometimes, partially. Clinical data generated for MDR can feed into a De Novo submission, but the FDA has its own views on trial design, endpoints, and statistical analysis. A clinical investigation designed only with MDR Annex XIV in mind can miss FDA-relevant elements. Design the clinical strategy with both regulators in mind from the beginning.

Do I need a US regulatory specialist for a De Novo? Realistically, yes. The pathway question alone benefits from specialist input, and the risk-and-controls argument at the core of a De Novo is harder to build without someone who has seen multiple De Novo submissions move through the agency. Tibor and Zechmeister Strategic Solutions do not hold themselves out as FDA experts; the US authority belongs to specialists who practice inside the FDA system daily.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices (MDR), in particular Article 2(1) (definition of medical device), Article 2(12) (intended purpose), Article 51 and Annex VIII (classification). Official Journal L 117, 5.5.2017, as amended, including Regulation (EU) 2023/607.
  2. U.S. Food and Drug Administration — Center for Devices and Radiological Health (CDRH), public guidance on the De Novo classification pathway, including the De Novo request process, risk-based classification, and special controls. Referenced at the general framework level. For current user fees, current guidance documents, current review performance, and the current status of De Novo in specific device categories, consult the FDA directly at https://www.fda.gov/medical-devices.
  3. EN ISO 13485:2016 + A11:2021 — Medical devices — Quality management systems — Requirements for regulatory purposes.

This post is part of the FDA & International Market Access series in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. A note on the limits of our expertise, stated once and clearly: Tibor's authority is in the EU MDR and the Notified Body system, and the FDA framing in this post stays deliberately at the general level. For specific De Novo questions on your device — pathway fit analysis, risk-and-controls argument development, pre-submission strategy, interaction with FDA review divisions, and 21 CFR Part 820 implementation detail — work with a US regulatory specialist who practices inside the FDA system daily. This post is the orientation. The specialist is the execution.