MDR Article 27 and Annex VI Part C require the UDI carrier to appear on the label of the device and on all higher levels of packaging. The carrier must present the UDI in two forms at once: machine-readable through an Automatic Identification and Data Capture (AIDC) technology. Typically a linear barcode or a GS1 DataMatrix. And human-readable (HRI) as plain characters that a person can read without a scanner. Both forms encode the same two components: the UDI-DI (device identifier, static per model and packaging level) and the applicable UDI-PI elements (lot, serial, software identification, manufacturing date, expiry date). The Basic UDI-DI is not printed on the label. Where the label is too small to carry both AIDC and HRI, the carrier may be placed on the next higher packaging layer under the space-constraint rules of Annex VI Part C.

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

TL;DR

  • MDR Article 27(3) requires the manufacturer to assign a UDI to the device and place the UDI carrier on the label of the device and on all higher levels of packaging before placing the device on the market.
  • Annex VI Part C specifies that the UDI carrier must present both an AIDC (machine-readable) representation and a human-readable interpretation of the UDI.
  • The UDI printed on the label is the full UDI. UDI-DI plus the applicable UDI-PI elements. The Basic UDI-DI stays in the technical documentation and the database; it is not on the label.
  • Shipping containers are out of scope: the carrier goes on the device label and on higher levels of packaging up to the sales or dispatch unit, but pure logistics containers are excluded by Annex VI Part C.
  • For reusable devices intended to be reprocessed, Annex VI Part C additionally requires the UDI to be directly marked on the device itself unless the exceptions in the Annex apply.

You have a UDI. Now where does it actually go?

A founder finishes the cluster posts on what a UDI is and how UDI-DI differs from UDI-PI, picks an issuing entity, generates the identifiers. And then hits a different question. Where on the physical label does this thing go, what does it look like, and how does a Notified Body auditor decide whether the label passes or fails?

The answer sits in two places and nowhere else. Article 27 establishes the obligation. Annex VI Part C specifies the carrier requirements, the dual AIDC/HRI rule, the small-device exception, the direct marking rule for reusables, and the change categories. This post walks through the display requirements in the order they matter for a label design review, connects each rule to the underlying text of the Regulation, and closes with the recurring mistakes we see in first-time audits.

Article 27 carrier requirements. What the Regulation actually mandates

MDR Article 27 establishes the UDI system, defines who assigns identifiers, and places the carrier obligation squarely on the manufacturer. Article 27(3) requires the manufacturer, before placing a device on the market, to assign a UDI to the device and, where applicable, to all higher levels of packaging. Article 27(4) then requires the UDI carrier to be placed on the label of the device and on all higher levels of packaging, with shipping containers excluded from that definition. (Regulation (EU) 2017/745, Article 27.)

"Higher levels of packaging" is the phrase that trips founders up. It does not mean the truck. It means every packaging level that is part of the sales or distribution configuration of the device. The unit blister, the inner carton, the shelf box, the dispatch unit as presented to the buyer. A pallet of cartons being moved by a logistics provider is a shipping container and is out of scope. A sales carton of twenty individually labelled units is in scope, and needs its own UDI carrier encoding its own UDI-DI for that packaging level.

The carrier itself is not specified to a single technology. Annex VI Part C recognises that the UDI carrier can be implemented with any Automatic Identification and Data Capture technology that correctly encodes the UDI under the chosen issuing entity's rules. Linear barcodes, two-dimensional matrix codes, RFID, and other AIDC formats are all permitted. In practice, the overwhelming default is a linear barcode (GS1-128 or equivalent) or a GS1 DataMatrix 2D code, because those are the formats hospital and distributor scanners actually read without special tooling.

AIDC and human-readable interpretation. Why both are required

Annex VI Part C sets out the dual-format rule that catches more first-time labels than any other. The UDI carrier must present the UDI in AIDC form. The machine-readable code that a scanner resolves. And in human-readable interpretation (HRI) form: the plain characters that a person can read off the label without a scanner.

The reason both are required is that the two forms serve two different readers. AIDC exists for the scanner at the distributor dock, the hospital stockroom, the operating theatre, the PMS data pipeline. HRI exists for the person who needs to read the identifier when the scanner is broken, the label is creased, the software is offline, or the incident report has to be filled in from a photograph. Neither form alone is enough. A label with only AIDC is unreadable when the scanner fails. A label with only HRI is not traceable at the speeds modern distribution requires.

The HRI must encode the same UDI elements as the AIDC. If the AIDC carries UDI-DI plus lot plus expiry, the HRI carries UDI-DI plus lot plus expiry. Omitting elements from the HRI because "the scanner has it anyway" is a finding. Presenting HRI elements in a format that a human cannot actually parse. Microscopic type, no labels on the fields, inconsistent ordering between sister products. Is also a finding, because the HRI exists to be humanly readable, not just humanly present.

The one concession Annex VI Part C makes is that where the label is physically too small to carry both AIDC and HRI, the carrier may be placed on the next higher level of packaging rather than on the immediate label. The concession is a space rule, not a convenience rule. "We wanted more room for branding" is not a valid invocation of the small-device exception.

UDI-DI vs UDI-PI on the label. What is encoded and what is not

The full UDI on the label is UDI-DI plus the applicable UDI-PI elements, encoded together inside a single carrier. The companion post on UDI-DI versus UDI-PI walks through the conceptual difference; the label-level consequences are the specific point here.

Inside a GS1 carrier, each element is marked by an Application Identifier (AI) that tells the parser what the following digits represent. AI (01) marks the GTIN that serves as the UDI-DI. AI (10) marks the lot or batch number. AI (21) marks the serial number. AI (11) marks the manufacturing date. AI (17) marks the expiry date. Software identifiers are carried with the AIs designated for that purpose under the current GS1 rules. HIBCC and the other designated issuing entities use different syntax conventions but the same underlying principle: each element of the UDI is tagged so the parser knows which part is UDI-DI and which is UDI-PI.

Two practical rules follow. First, the HRI must display the elements in a way that makes the tagging visible to the human eye. Typically with the AI in parentheses before each field, or with labelled captions ("LOT:", "EXP:", "SN:") so the reader knows what each number is. Second, the UDI-DI and the UDI-PI are two different things with two different lifecycles: the UDI-DI is static for the whole production run of a model at that packaging level, and the UDI-PI changes with every new lot, serial, software release, or date. The print system must treat the UDI-DI as a fixed element in the label artwork and the UDI-PI as a variable element populated at the point of labelling. Mixing the two up. Hard-coding a lot number into the template, or dynamically generating a UDI-DI that should have been stable. Is a common failure mode.

What is not on the label is the Basic UDI-DI. The Basic UDI-DI is defined in MDR Article 2(15) as the primary access key for the UDI database and the identifier used on the EU declaration of conformity, in the technical documentation, and on the Notified Body's certificate. It is administrative. It does not belong in the carrier or in the HRI on the physical label. Printing it there is not a compliance violation in itself but it consistently signals to an auditor that the manufacturer has not fully understood the identifier hierarchy.

The small-device exception. When the carrier can move up a level

Annex VI Part C permits a manufacturer, where the physical label is too small to accommodate both the AIDC and the HRI of the UDI, to place the carrier on the next higher packaging level instead of on the immediate device label. The exception is narrow and has to be justified.

Three things follow from the wording. The exception is about physical space, not about design preference. The burden is on the manufacturer to demonstrate that the label is genuinely too small. The exception moves the carrier up one level; it does not eliminate the requirement. And the exception is evaluated per label, not per device family. A given SKU may invoke it while a larger sibling SKU does not.

For implants, micro-components, and single-use disposables with tiny primary packaging, the small-device exception is the usual operating mode. The UDI carrier lives on the sterile pouch or the immediate sales unit, not on the device itself. The immediate device label still carries the information from Annex I Chapter III Section 23.2 that must stay with the device at the point of use. The companion post on MDR labelling requirements walks through that layered hierarchy. But the UDI carrier specifically may be lifted to the next layer under Annex VI Part C.

For reusable devices, Annex VI Part C goes the other direction. In addition to the carrier on the label and on higher packaging, the UDI must be directly marked on the device itself so that it survives cleaning and reprocessing and can be read on the individual unit throughout its service life. Exceptions to direct marking are set out in the Annex for cases where marking is not technologically feasible or would compromise the safety or performance of the device, and those exceptions have to be justified in the technical documentation.

The packaging hierarchy. Which UDI goes on which layer

Each packaging level has its own UDI-DI, and each level's carrier encodes that level's UDI-DI plus the applicable UDI-PI. This is the rule that founders most often want to skip, because assigning a separate UDI-DI for the unit and for the box "feels redundant." It is not redundant. It is the only way that a scan at a given point in the supply chain returns an identifier for the object actually being scanned.

Walk a single-use sterile device through its layers. The sterile pouch is the immediate sales unit and carries one UDI-DI for "one pouch of this device." A carton containing twenty pouches carries a different UDI-DI for "carton of twenty of this device." A shelf box containing five cartons, if that level is part of the sales hierarchy, carries its own UDI-DI again. Each carrier on each layer encodes the UDI-DI for that layer and the UDI-PI elements that apply to the production run being packaged. Shipping containers above the sales hierarchy. The logistics pallet, the freight box. Are excluded under Article 27(4).

The practical consequence for label design is that the label artwork is not one file but a family of files, one per packaging level, each with its own UDI-DI hard-coded and its own UDI-PI field variables populated at print time. Mixing the levels. Printing the unit UDI-DI on the carton carrier, or vice versa. Is a finding that shows up fast in distributor reconciliation and eventually in a vigilance investigation.

Common mistakes on UDI labels

A small set of errors recurs on almost every first UDI label we review.

AIDC without HRI, or HRI without AIDC. Both forms are required by Annex VI Part C. Missing either is a finding.

HRI that omits elements the AIDC carries. If the carrier encodes lot and expiry, the HRI must display lot and expiry. Minimal HRI that shows only the UDI-DI is non-compliant.

Printing the Basic UDI-DI on the label. The Basic UDI-DI lives in documentation and database, not on the physical label. It is harmless in itself but signals a misunderstanding.

Hard-coding the UDI-PI into the label template. The UDI-PI is variable at print time. Templates that bake in last year's lot number reliably produce mismatched labels in the next production run.

Using the same UDI-DI across different packaging levels. Each packaging level needs its own UDI-DI. Scanners at the carton dock need to see "carton," not "unit."

Invented symbols next to the UDI carrier. Symbols around the carrier must come from EN ISO 15223-1:2021. Manufacturer, expiry, lot, serial, and so on. And cannot be replaced with a custom icon because the brand team prefers it.

Claiming the small-device exception for design reasons. The exception is for physical space. "We wanted more room for the logo" does not qualify.

Forgetting direct marking on reusables. Reusable devices intended to be reprocessed must carry the UDI directly on the device under Annex VI Part C, subject to the Annex's exceptions. A label-only approach is incomplete.

The Subtract to Ship angle on UDI labelling

UDI display on the label is another clean application of the Subtract to Ship framework for MDR. The work that survives subtraction traces to Article 27 and Annex VI Part C and nothing else.

The real work is: generate UDI-DIs for every packaging level under the chosen issuing entity's rules; decide which UDI-PI elements apply for each device; design one label artwork file per packaging level with the carrier placed visibly, the AIDC and the HRI both present, and the human-readable fields labelled so a person can parse them; lock the UDI-DI as a static element and the UDI-PI as a variable element populated at print; add the direct marking for reusables; and run a small-device-exception decision for any SKU where the immediate label is physically too small for both forms. That is the entire UDI labelling workstream.

Everything else is a candidate for removal. Parallel barcode schemes for internal tracking that duplicate the UDI carrier, speculative RFID layers that are not required by the issuing entity or the Regulation, custom font work on the HRI that makes it harder to read, and "label readiness workshops" that do not end in a concrete template per packaging level. All of these come out. If a proposed activity does not trace to Article 27, Annex VI Part C, the issuing entity's encoding rules, or EN ISO 15223-1:2021 for the surrounding symbols, it does not belong in the plan.

Reality Check. Is your UDI label ready?

  1. Does every label carry the UDI in both AIDC form and HRI form, with both encoding the same UDI-DI and UDI-PI elements?
  2. Is the HRI laid out so a human can parse each field. With Application Identifiers in parentheses, or with LOT/SN/EXP captions. And not just present as an undifferentiated string of digits?
  3. Does each packaging level in your sales hierarchy have its own UDI-DI, and does each level's label carrier encode that level's UDI-DI?
  4. Is the UDI-DI locked as a static element in the label artwork, and is the UDI-PI populated as a variable at print time from the production system?
  5. For any SKU where the immediate label is physically too small, have you documented the small-device exception and confirmed the carrier sits on the next higher packaging level?
  6. For any reusable device, is the UDI directly marked on the device itself in a form that survives reprocessing, or is the exception to direct marking documented in the technical file?
  7. Are the symbols around the UDI carrier taken from the current edition of EN ISO 15223-1:2021, and is the Basic UDI-DI kept off the label?
  8. Do the UDI-DIs printed on your labels match exactly the UDI-DIs submitted to the UDI database in EUDAMED under Article 29?

If you cannot answer five or more of these cleanly, the UDI label is not yet ready for a Notified Body audit.

Frequently Asked Questions

Where does the UDI have to appear on a medical device label? MDR Article 27(4) requires the UDI carrier to be placed on the label of the device and on all higher levels of packaging, excluding shipping containers. Annex VI Part C specifies that the carrier must present both an AIDC (machine-readable) form and a human-readable interpretation of the UDI.

Does the UDI on the label need to be both a barcode and readable text? Yes. Annex VI Part C requires both forms: the AIDC representation. Typically a linear barcode or a 2D data matrix. And the human-readable interpretation in plain characters. Both must encode the same UDI-DI and the same applicable UDI-PI elements.

What happens if my device label is too small for both the barcode and the readable text? Annex VI Part C allows the UDI carrier to be placed on the next higher packaging level where the immediate label is physically too small to accommodate both the AIDC and the human-readable interpretation. The exception is about physical space and has to be justified in the technical documentation.

Is the Basic UDI-DI printed on the label? No. The Basic UDI-DI is defined in MDR Article 2(15) as the primary access key for the UDI database and appears in the technical documentation, on the declaration of conformity, and on the Notified Body certificate. It is not carried on the physical label. What is carried on the label is the full UDI. UDI-DI plus the applicable UDI-PI elements.

Does each packaging level need its own UDI? Yes. Each level of the sales and distribution packaging hierarchy. Unit, inner carton, sales carton, dispatch unit. Needs its own UDI-DI, encoded in its own carrier on that level's label. Shipping containers used purely for logistics are excluded under Article 27(4).

Do reusable devices need the UDI on the device itself? Yes. Annex VI Part C requires direct marking of the UDI on reusable devices intended to be reprocessed, in addition to the carrier on the label and on higher packaging levels. Exceptions to direct marking are set out in the Annex and must be justified where invoked.

Which AIDC format should I use. Linear barcode or 2D data matrix? Either is permitted by Annex VI Part C, provided the format correctly encodes the UDI under the rules of the chosen issuing entity. In practice, GS1 DataMatrix is increasingly preferred for small labels because it encodes more data in less space, while GS1-128 linear barcodes remain common for larger labels. The choice should align with distributor and hospital scanning infrastructure in the target market.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Article 27 (Unique Device Identification system. Assignment of the UDI, placement of the UDI carrier on the label of the device and on all higher levels of packaging), Article 2(15) (definition of Basic UDI-DI), and Annex VI Part C (the UDI system, including the AIDC and human-readable interpretation requirements, the small-device exception, direct marking of reusable devices, and change categories triggering a new UDI-DI). Official Journal L 117, 5.5.2017.
  2. Commission Implementing Regulation (EU) 2021/2078 of 26 November 2021 laying down rules for the application of Regulation (EU) 2017/745 of the European Parliament and of the Council as regards the European Database on Medical Devices (Eudamed). OJ L 426, 29.11.2021.
  3. EN ISO 15223-1:2021, "Medical devices. Symbols to be used with medical device labels, labelling and information to be supplied. Part 1: General requirements."

This post is part of the Technical Documentation & Labeling cluster in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. A UDI label that passes audit is not the product of design instinct; it is the product of reading Article 27 and Annex VI Part C in full, placing the carrier where the Regulation says it belongs, and letting the encoding rules of the chosen issuing entity decide every character inside the carrier.