In the United States, section 513 of the Federal Food, Drug, and Cosmetic Act sets a risk-based system of three classes, and devices of a type that was not on the market before the Medical Device Amendments of 28 May 1976 are placed in Class III automatically unless FDA classifies them otherwise (FDA, Regulatory Controls; FDA, Reclassification). In the European Union, Regulation (EU) 2017/745 (MDR) has applied since 26 May 2021 and divides devices into classes I, IIa, IIb and III under Article 51, using the 22 rules in Annex VIII (MDR; MDCG 2021-24 Rev.1). Both systems sort devices by risk, but they reach a class in very different ways.
The consequence is that the classification of medical devices cannot be copied from one market to the other. In this post we explain how each system decides the class, where the two diverge, and what that means for a manufacturer, in particular a software manufacturer, planning for both markets.
Two different questions
The US system asks which existing device type a product belongs to. FDA has established classifications for approximately 1,700 generic types of devices, grouped into 16 medical specialty panels in 21 CFR Parts 862 to 892, and each type is assigned to Class I, II or III (FDA, Classify Your Medical Device). The manufacturer’s task is to find the classification regulation whose description matches the device.
The EU system asks which rules apply to this particular device. Under Article 51 of the MDR, devices are divided into classes I, IIa, IIb and III, “taking into account the intended purpose of the devices and their inherent risks”, and classification is carried out in accordance with Annex VIII (MDR). The manufacturer applies the rules to its own device and documents the result.
In both systems the starting point is the same: what the manufacturer claims the device is for. FDA states that classification depends on the intended use of the device and on its indications for use (FDA); under the MDR, the rules are applied to the intended purpose (MDR, Art. 51). A change in claims can therefore change the class in either market, without a single line of code or a single component changing.
The United States: classification by device type
FDA describes three classes, each with its own set of controls (FDA, Regulatory Controls):
| Class | Risk as FDA describes it | Controls | Usual premarket route |
|---|---|---|---|
| Class I | Low to moderate | General controls | Exempt from 510(k) for most Class I devices; otherwise 510(k) |
| Class II | Moderate to high | General controls and special controls | 510(k), unless exempt |
| Class III | High | General controls and premarket approval (PMA) | PMA |
Sources: FDA, Regulatory Controls, Classify Your Medical Device and Class I / II Exemptions.
FDA is explicit that if a device is Class I or II and not exempt, a 510(k) is required, and that Class III devices need a PMA unless they are preamendments devices for which PMAs have not been called for (FDA). Exemptions are stated in the classification regulation itself, together with the limitations set out in 21 CFR 862.9 to 892.9, and an exempt device still has to meet the other regulatory controls unless the regulation expressly exempts it (FDA, Class I / II Exemptions).
In practice, the class is found by locating the right classification regulation. FDA points manufacturers to two routes: searching the Product Classification Database for part of the device name, or going through the listing for the relevant specialty panel in the CFR. Each classified device type has a seven-digit regulation number and a three-letter product code; FDA’s own example is the clinical mercury thermometer, 21 CFR 880.2920, product code FLK (FDA). The database is the index; the regulation is the legal basis.
When no existing device type fits, the device is not unclassified. A postamendments device, one not in commercial distribution before 28 May 1976, is classified into Class III automatically under section 513(f)(1), and remains there unless it is classified through the De Novo process under section 513(f)(2), reclassified, or found substantially equivalent to a predicate that does not require PMA (FDA, Reclassification). The De Novo pathway is meant for novel devices for which general controls, or general and special controls, provide reasonable assurance of safety and effectiveness, and a granted request creates a new classification regulation that later devices can use as a predicate. Where the classification is unclear, a manufacturer can ask FDA for a formal answer through a 513(g) request, which carries a user fee (FDA).
The European Union: classification by rules
Annex VIII of the MDR contains 22 classification rules: rules 1 to 4 for non-invasive devices, rules 5 to 8 for invasive devices, rules 9 to 13 for active devices and rules 14 to 22 as special rules (MDCG 2021-24 Rev.1). The rules work with defined concepts such as duration of use, invasiveness and active devices, which is why the definitions in Annex VIII need to be read before the rules themselves.
Two implementing rules decide how the result is reached. Where several rules, or several sub-rules within the same rule, apply to the same device, “the strictest rule and sub-rule resulting in the higher classification shall apply” (Annex VIII, Section 3.5). And software that drives a device or influences the use of a device falls within the same class as that device, while software that is independent of any other device is classified in its own right (Annex VIII, Section 3.3) (MDCG 2019-11 Rev.1). A classification rationale that only addresses the rule the manufacturer prefers is therefore incomplete; it has to show why none of the other rules leads to a higher class.
For software, Rule 11 is the rule that matters most. It reads, in summary of its three parts (MDCG 2019-11 Rev.1):
-
Software intended to provide information which is used to take decisions with diagnosis or therapeutic purposes is in class IIa, except where such decisions have an impact that may cause death or an irreversible deterioration of a person’s state of health (class III), or a serious deterioration of a person’s state of health or a surgical intervention (class IIb).
-
Software intended to monitor physiological processes is in class IIa, except where it is intended for monitoring vital physiological parameters and the nature of variations of those parameters could result in immediate danger to the patient (class IIb).
-
All other software is in class I.
Rule 11 makes the clinical consequence of the information, not the technology, the deciding factor. That is also why most software that informs diagnostic or therapeutic decisions ends up in class IIa or higher under the MDR, a point we discussed in more detail in our article on why software cannot escape CE marking.
The class then determines the conformity assessment route and the involvement of a notified body. If manufacturer and notified body disagree on the classification, the MDR refers the matter to the competent authority of the Member State where the manufacturer has its registered place of business (MDCG 2021-24 Rev.1). In vitro diagnostic devices follow a separate system altogether: under Article 47 of Regulation (EU) 2017/746 (IVDR), they are divided into classes A, B, C and D according to Annex VIII of that regulation (IVDR).
Classification of medical devices: where the EU and US diverge
| Topic | EU (MDR) | US (FDA) |
|---|---|---|
| Classes | I, IIa, IIb, III (Art. 51) | I, II, III (section 513) |
| How the class is decided | Manufacturer applies the Annex VIII rules to its own device | Device type is matched to an existing classification regulation in 21 CFR 862 to 892 |
| Several rules apply | Strictest rule and sub-rule apply (Annex VIII, Section 3.5) | Not applicable; one classification regulation per device type |
| Software | Rule 11, plus implementing rule 3.3 | Same device-type regulations as other devices; no separate software rule |
| No match | Not applicable; the rules always produce a class | Automatically Class III (513(f)(1)); De Novo, reclassification or PMA |
| Formal determination | Competent authority in case of dispute with the notified body | 513(g) request for information |
| IVDs | Separate regulation, classes A to D (IVDR Art. 47) | Classified within the same system, in the relevant panels |
Sources: MDR, IVDR, MDCG 2021-24 Rev.1, MDCG 2019-11 Rev.1; FDA, Classify Your Medical Device, Device Classification Panels and Reclassification.
The most common mistake is to treat the numbers as equivalent. There is no one-to-one mapping between an EU class and a US class, and a Class II device in the US is not the same thing as a class IIa or IIb device in the EU. Each classification needs its own rationale, written against its own legal test.
Example: one product, two versions
To show how this works in practice, take a hypothetical case. A software manufacturer has two versions of the same product: one that displays and stores medical images, and one that also flags suspected findings for a clinician to review.
In the EU, the manufacturer first establishes whether each version qualifies as medical device software, following MDCG 2019-11 Rev.1, and then applies Rule 11 together with any other rule that might apply. The second version provides information used to take decisions with diagnostic purposes, so Rule 11 places it in class IIa at least; whether it ends up in class IIb or III depends on the impact of the decisions it informs, and that analysis has to be written down, not assumed.
In the US, the manufacturer compares the intended use and indications for use of each version with the identification text of candidate classification regulations. The two versions may well fall under different regulations, because their claims differ. If no regulation matches the second version and there is no predicate, it is a postamendments device in Class III by operation of law, and the realistic routes become De Novo or PMA. If the answer is not clear, a 513(g) request gives FDA’s view before the submission strategy is fixed.
The point of the example is that the claims, not the code, decide the class in both markets, and that the two analyses use different questions. A team that writes one rationale and translates it will usually find the gap at the worst moment: during notified body review or in an FDA deficiency letter.
Keeping classification under control
Classification is not a one-time decision. In the US, the class of a device type can change through reclassification under section 513(e), or under section 513(f)(3) for postamendments devices automatically placed in Class III, and reclassification applies to a device type rather than an individual device (FDA, Reclassification). In the EU, the class follows the intended purpose, so a new indication, a new user group or a new output can move a device to another rule or sub-rule. In both cases, the classification rationale should sit under change control together with the intended purpose and the technical documentation.
The rules themselves may also change. On 16 December 2025 the European Commission published a proposal to amend the MDR and IVDR, COM(2025) 1023 (European Commission). It is a proposal, not law, and the classification rules in Annex VIII continue to apply as they stand.
Conclusion
The EU and the US both classify medical devices by risk, but they ask different questions. FDA matches a device to an existing device type and its regulation, with automatic Class III and De Novo as the fallback when nothing fits. The MDR has the manufacturer apply 22 rules to its own device, with the strictest applicable rule deciding the class and Rule 11 setting the terms for software. In our view, the safest approach is to treat them as two separate analyses from the start, both anchored in the same carefully written intended purpose.
Our advice is simple: settle the intended purpose first, classify for each market separately, keep the rationale under change control, and revisit it every time the claims change. The detailed rules for each system deserve their own reading, which is why we cover the EU classification rules and FDA classification in separate articles. For a wider view of how classification feeds into market entry, see our article on regulatory strategy for the EU and US, and for help building a classification rationale that holds up in both markets, our regulatory strategy team can support you. If you want to discuss your device, contact the MedQAIR team.
Frequently asked questions
Is an FDA Class II device the same as an EU class IIa or IIb device?
No. The two systems use different legal tests: FDA assigns a class to a device type through a classification regulation, while the MDR applies the Annex VIII rules to the individual device. There is no one-to-one mapping between the classes, so each market needs its own rationale (FDA; MDR, Art. 51).
Who decides the class of a medical device?
In the EU, the manufacturer classifies the device under Annex VIII; disagreements with the notified body go to the competent authority (MDCG 2021-24 Rev.1). In the US, the class is set by FDA’s classification regulation for the device type, and a manufacturer can request FDA’s view through a 513(g) request (FDA).
What happens if my device does not match any FDA classification regulation?
A device of a type not marketed before 28 May 1976 is automatically Class III under section 513(f)(1). It can move to Class I or II through a De Novo request, reclassification, or a substantial equivalence finding against a predicate that does not require PMA (FDA, Reclassification).
If several MDR rules apply to my device, which one counts?
The strictest rule and sub-rule resulting in the higher classification apply (Annex VIII, Section 3.5). The rationale should show that all potentially relevant rules were considered (MDCG 2019-11 Rev.1).
Are IVDs classified under the same rules?
Not in the EU. In vitro diagnostic devices fall under Regulation (EU) 2017/746, which uses classes A, B, C and D under Article 47 and its own Annex VIII (IVDR).
