IVDR classification decides almost everything downstream: whether a notified body is involved at all, which conformity assessment route applies, whether an EU reference laboratory has a veto over your certificate, how often a performance evaluation report has to be updated, and which transitional deadline your legacy device falls under.
It is also where the in vitro diagnostic regime departs most sharply from the medical device one. Regulation (EU) 2017/746 uses four classes and seven rules; Regulation (EU) 2017/745 uses four different classes and twenty-two rules. Nothing transfers.
Why classification changed everything for IVDs
Directive 98/79/EC worked from lists. Annex II List A and List B named specific device types; self-testing devices had their own regime; and everything else — the large majority of IVDs on the market — was self-declared by the manufacturer without a notified body.
The IVDR replaced that with risk-based rules. Under Annex VIII every device sorts into class A, B, C or D, and only class A devices that are not placed on the market sterile are self-declared. Devices that lawfully carried a CE mark under the Directive without anyone else looking at them now need a notified body.
That shift is why Article 110(3b) exists as a separate transitional bracket, and why notified body capacity has been the sector’s binding constraint ever since.
The implementing rules come first
Annex VIII Section 1 sets ten implementing rules that govern how the seven classification rules are applied. They decide more borderline cases than the rules themselves.
| Rule | What it does |
|---|---|
| 1.1 | Application of the classification rules is governed by the intended purpose of the device |
| 1.2 | Where a device is used in combination with another, the rules apply separately to each |
| 1.3 | Accessories are classified in their own right, separately from the device they are used with |
| 1.4 | Software that drives a device or influences its use falls in the same class as that device. Software independent of any other device is classified in its own right |
| 1.5 | Calibrators intended to be used with a device take the same class as the device |
| 1.6 | Control materials with assigned values, for one or multiple analytes, take the same class as the device |
| 1.7 | The manufacturer shall take all classification and implementing rules into consideration |
| 1.8 | Where multiple intended purposes put a device in more than one class, it takes the higher class |
| 1.9 | Where several rules apply, the rule resulting in the higher classification applies |
| 1.10 | Each rule applies to first line assays, confirmatory assays and supplemental assays |
Rule 1.1 is the one to internalise. Classification follows the intended purpose, not the technology. The same assay on the same instrument can be class B or class D depending only on what it is claimed for — which means a marketing decision to broaden a claim is a regulatory decision about class.
Rules 1.8 and 1.9 are why a classification record needs to show the rules you rejected, not only the one you applied. A reviewer will ask why a higher rule did not apply, and “we chose Rule 6” is not an answer.
Rule 1 — class D
Devices intended for any of the following are class D:
- detection of the presence of, or exposure to, a transmissible agent in blood, blood components, cells, tissues or organs, or any of their derivatives, to assess their suitability for transfusion, transplantation or cell administration;
- detection of the presence of, or exposure to, a transmissible agent that causes a life-threatening disease with a high or suspected high risk of propagation;
- determining the infectious load of a life-threatening disease where monitoring is critical in the process of patient management.
Class D is the only class that attracts EU reference laboratory verification under Article 48(5) and ongoing batch verification under Annex IX Section 4.12. Establishing it late compresses a programme that has an external laboratory in its critical path.
Rule 2 — blood grouping and tissue typing
Devices for blood grouping, for determining foeto-maternal blood group incompatibility, or for tissue typing to ensure immunological compatibility of blood, blood components, cells, tissue or organs intended for transfusion, transplantation or cell administration are class C.
Except where they determine any of these markers, in which case they are class D:
- ABO system [A (ABO1), B (ABO2), AB (ABO3)];
- Rhesus system [RH1 (D), RHW1, RH2 (C), RH3 (E), RH4 (c), RH5 (e)];
- Kell system [Kel1 (K)];
- Kidd system [JK1 (Jka), JK2 (Jkb)];
- Duffy system [FY1 (Fya), FY2 (Fyb)].
The named-marker list is exhaustive and worth reading literally. A blood grouping device outside it sits at class C.
Rule 3 — class C, and thirteen ways to get there
Rule 3 is the longest rule and the one most devices land on. Devices are class C if intended:
- for detecting the presence of, or exposure to, a sexually transmitted agent;
- for detecting the presence in cerebrospinal fluid or blood of an infectious agent without a high or suspected high risk of propagation;
- for detecting the presence of an infectious agent where there is a significant risk that an erroneous result would cause death or severe disability to the individual, foetus or embryo being tested, or to the individual’s offspring;
- for pre-natal screening of women to determine their immune status towards transmissible agents;
- for determining infective disease status or immune status, where an erroneous result would lead to a patient management decision resulting in a life-threatening situation;
- to be used as companion diagnostics;
- for disease staging, where an erroneous result would lead to a patient management decision resulting in a life-threatening situation;
- for screening, diagnosis or staging of cancer;
- for human genetic testing;
- for monitoring levels of medicinal products, substances or biological components, where an erroneous result would lead to a patient management decision resulting in a life-threatening situation;
- for management of patients suffering from a life-threatening disease or condition;
- for screening for congenital disorders in the embryo or foetus;
- for screening for congenital disorders in new-born babies where failure to detect and treat could lead to life-threatening situations or severe disabilities.
Points (h) and (i) are broad and unqualified. Any device for screening, diagnosis or staging of cancer is class C, with no erroneous-result test attached. Any human genetic testing device is class C. Neither depends on severity.
Several of the others turn on a judgement — “significant risk that an erroneous result would cause death or severe disability”, “a patient management decision resulting in a life-threatening situation”. Those are the rows where a classification record has to reason rather than assert, and where a device sitting at class B on an optimistic reading is most exposed.
Companion diagnostics
Rule 3(f) puts companion diagnostics at class C regardless of analyte or specimen. The class is only half the consequence: under Article 48(7) a class C companion diagnostic goes through the Annex IX Section 5.2 technical documentation assessment for every device rather than on a representative basis, and the notified body must consult a medicines competent authority designated under Directive 2001/83/EC, or the EMA.
Rule 4 — self-testing and near-patient testing
(a) Devices intended for self-testing are class C — except devices for the detection of pregnancy, for fertility testing, and for determining cholesterol level, and devices for the detection of glucose, erythrocytes, leucocytes and bacteria in urine, which are class B.
(b) Devices intended for near-patient testing are classified in their own right.
Rule 4(b) is frequently misread as a downgrade or an exemption. It is neither. It means near-patient testing does not itself set a class — you classify the device on its own intended purpose under the other rules. What near-patient status does change is the route: at class B, C or D it adds the Annex IX Section 5.1 technical documentation assessment, which requires results of studies carried out with intended users.
That connects to Annex I Section 9.4, which requires performance to be specifically checked where it may be affected in normal use: for self-testing, the performance obtained by laypersons; for near-patient testing, the performance obtained in relevant environments such as the patient’s home, emergency units and ambulances. That is a study, not an assertion.
Rule 5 — class A
The following are class A:
- (a) products for general laboratory use, accessories which possess no critical characteristics, buffer solutions, washing solutions, and general culture media and histological stains, intended by the manufacturer to make them suitable for in vitro diagnostic procedures relating to a specific examination;
- (b) instruments intended by the manufacturer specifically to be used for in vitro diagnostic procedures;
- (c) specimen receptacles.
Class A is the only self-declared class — and only when the device is not placed on the market in sterile condition. Under Article 48(10), a class A device placed on the market sterile requires Annex IX or Annex XI, with notified body involvement limited to establishing, securing and maintaining sterile conditions. The class is still A; the route is not self-declaration.
Class A also carries the last UDI carrier deadline still in the future: Article 113(3)(e) applies the Article 24(4) carrier obligation to class A devices from 26 May 2027.
Rules 6 and 7 — the residual classes
Rule 6: devices not covered by the above rules are class B.
Rule 7: devices which are controls without a quantitative or qualitative assigned value are class B.
Rule 7 pairs with implementing rule 1.6: a control material with an assigned value takes the class of the device it is intended for; a control without one is class B in its own right.
Rule 6 is where most routine clinical chemistry and haematology assays land. It is a residual rule, not a default — reaching it means having considered and rejected Rules 1 to 5, and that reasoning belongs in the record.
What the class then determines
| Class | Conformity assessment | Post-market reporting | SSP |
|---|---|---|---|
| D | Annex IX Ch. I, II (except Sec. 5) and III, or Annex X with Annex XI. EU reference laboratory verification and batch verification | PSUR at least annually, submitted via the Article 87 system for notified body review | Yes |
| C | Annex IX Ch. I and III plus technical documentation assessment for one representative device per generic device group | PSUR at least annually, made available to the notified body | Yes |
| B | Annex IX Ch. I and III plus technical documentation assessment for one representative device per category of devices | Post-market surveillance report, updated when necessary | No |
| A | Self-declared on Annexes II and III technical documentation — sterility only if sterile | Post-market surveillance report, updated when necessary | No |
The class also fixes the performance evaluation cycle: under Article 56(6), class C and class D performance evaluation reports must be updated when necessary but at least annually.
And it fixes the transitional bracket for legacy devices — class D to 31 December 2027, class C to 2028, class B and class A sterile to 2029 — with condition deadlines falling years earlier. See IVDR transition deadlines.
What a classification record should contain
A defensible record is per device and per variant, and it shows working:
- the intended purpose, quoted exactly as it appears in the technical documentation — not paraphrased;
- the rule applied, to the sub-point where the rule has them;
- the implementing rule relied on where more than one classification rule could apply;
- the rules considered and rejected, each with a reason and the class it would have produced;
- the resulting class, and who approved it;
- the consequences that follow — route, overlays, reporting cycle, transitional bracket.
Annex II Section 1.1(f) requires the technical documentation to contain the risk class and the justification for the classification rules applied. A class with no reasoning behind it does not meet that.
Review the determination on any change of intended purpose, analyte, specimen type or testing population. A change of intended purpose is a classification event before it is a labelling one — and for a legacy device it may also be the significant change that ends the Article 110 extension.
Our EU IVDR Toolkit includes a device classification procedure, a qualification and borderline procedure, a companion diagnostics guide and a classification rationale record built around that structure. For the wider picture see our guide to the EU IVDR, and for what the class determines about your evidence, IVDR performance evaluation.
Frequently asked questions
How many classification rules does the IVDR have?
Seven, in Annex VIII Section 2, preceded by ten implementing rules in Section 1. That is quite different from Regulation (EU) 2017/745, which uses twenty-two rules and the classes I, IIa, IIb and III.
What are the IVDR device classes?
Class A, B, C and D, with D the highest risk. Only class A devices not placed on the market in sterile condition are self-declared; all other classes require notified body involvement.
What class is a companion diagnostic?
Class C, under Annex VIII Rule 3(f). It also attracts the Annex IX Section 5.2 procedure for every device, including consultation with a medicines competent authority or the EMA.
Are self-tests always class C?
No. Rule 4(a) makes self-testing devices class C by default, but names exceptions at class B: pregnancy detection, fertility testing, cholesterol determination, and detection of glucose, erythrocytes, leucocytes and bacteria in urine.
Does near-patient testing change the class?
No. Rule 4(b) says near-patient testing devices are classified in their own right, so the class comes from the other rules. It changes the route instead: at class B, C or D it adds the Annex IX Section 5.1 technical documentation assessment, which requires studies carried out with intended users.
What happens if two classification rules apply?
Implementing rule 1.9 applies the rule resulting in the higher classification. Implementing rule 1.8 does the same where multiple intended purposes place a device in more than one class.
How are accessories and software classified?
Accessories are classified in their own right, separately from the device they are used with (implementing rule 1.3). Software that drives a device or influences its use takes the same class as that device; software independent of any other device is classified in its own right (implementing rule 1.4).