UDI marking requires most medical devices to carry a Unique Device Identifier in two forms on the label and package — plain text a person can read and an AIDC (Automatic Identification and Data Capture) barcode a scanner can read — under the FDA UDI rule codified in 21 CFR Part 830 and 21 CFR Part 801. The identifier combines a static UDI-DI with a variable UDI-PI. Choose a printed label for single-use devices; if a device is used more than once and reprocessed between uses, 21 CFR 801.45 also requires a permanent direct mark — commonly a laser-etched Data Matrix that survives repeated sterilization.

For an OEM engineering or quality team, the real deliverable is a marking specification: which identifier, which data carrier, which method, and what verification proves the mark still scans at end of life. The sections below cover what the rule requires and where the label-versus-laser decision gets made.

What the FDA UDI rule actually requires

The U.S. UDI system was established by a 2013 FDA final rule. The system requirements live in 21 CFR Part 830, and the labeling amendments sit in 21 CFR Part 801; with narrow exceptions under 21 CFR 801.30, most devices must bear a UDI on the label and package. The goal is a single standardized identifier that ties a physical device back to a database record and its production history. FDA's UDI Basics is the plain-language starting point.

What must appear on a device label?

Every UDI must appear in two forms on the label and package: a plain-text, human-readable version and an AIDC (Automatic Identification and Data Capture) version, which is the barcode. A barcode alone does not comply, and neither does human-readable text alone; the rule requires both together. The practical takeaway: any marking spec has to call out both forms explicitly, or the device fails at inspection.

Who has to comply with UDI marking, and by when?

UDI requirements phased in by device class and are now fully in effect. Class III devices had a September 24, 2014 label deadline; implantable, life-supporting, and life-sustaining devices, September 24, 2015; Class II, September 24, 2016; and Class I and unclassified devices, originally September 24, 2018, with FDA enforcement discretion extending compliance to September 24, 2022 (GUDID submission for those devices to December 8, 2022). Direct-mark deadlines generally fell about two years after each class's label deadline. By 2026, all of these dates have passed — UDI is a baseline, not a future project.

UDI-DI vs. UDI-PI: the two parts of the code

A UDI is not one number. It is two segments with different owners and behaviors, and understanding the split is what keeps a marking program from breaking down at the OEM and contract-manufacturer boundary.

What is the difference between the UDI-DI and the UDI-PI?

The UDI-DI (Device Identifier) is mandatory and static: it identifies the labeler and the specific version or model, and it does not change from unit to unit. The UDI-PI (Production Identifier) is conditional and variable: it conveys the production data present on the label, such as lot or batch number, serial number, expiration date, and manufacturing date. The DI is assigned once, up front; the PI is generated per unit or per lot on the line. That is why traceability matters — the PI is only as trustworthy as the records behind it, as our ISO 13485 PCB assembly compliance checklist details.

Who assigns the UDI, and where does GUDID fit in?

FDA does not issue UDIs. It accredits three issuing agencies — GS1, HIBCC, and ICCBBA — whose systems the labeler uses to obtain a Device Identifier, for example a GS1 GTIN. The labeler, usually the OEM or specification owner, then submits the DI record to FDA's Global Unique Device Identification Database (GUDID); the public searches those records through AccessGUDID. Note: GUDID stores only the DI and its attributes; the PI (lot, serial, expiration, manufacturing date) is never submitted to it. Assignment and submission belong to the labeler, not the contract manufacturer.

Label vs. direct part marking: choosing your UDI approach

Which devices must carry a direct part mark?

Under 21 CFR 801.45, a device that must bear a UDI on its label must also carry a permanent UDI marked directly on the device if it is used more than once and reprocessed — cleaned, disinfected, and/or sterilized — before each use. Single-use devices do not need a direct mark; a compliant label on the device and packaging is enough. FDA guidance is explicit that devices only cleaned between uses are not reprocessed before each use; see the eCFR text of 21 CFR 801.45. So the first design question is simply: is this device reusable and reprocessed? If yes, plan permanent marking.

When should you use a 1D barcode vs. a 2D Data Matrix?

Data carriers fall into two families: linear, 1D barcodes such as GS1-128, and 2D symbols, principally the Data Matrix ECC 200. Data Matrix is compact, high-density, omnidirectional, and Reed-Solomon error-corrected, so a partially damaged symbol can still decode — which is why it is the default for small devices and for direct part marking. Match the carrier and cell size to the available surface: a symbol too dense for the space, or too fine to resolve, grades poorly and fails scanning. As a rule, packaging can carry 1D; small reusable instruments almost always need 2D.

Why is laser marking the default for reusable instruments?

Laser marking is favored for direct part marking because it is permanent and survives repeated autoclave and reprocessing cycles without washing off. But laser is not legally mandated: the rule requires a permanent, durable mark that survives the device's service life, and electrochemical etching, dot-peen, and durable ink can also qualify. Whatever the method, it must be validated for the specific substrate — stainless steel, titanium, aluminum, and polymer all behave differently under a laser. Laser plus a Data Matrix is simply the most common combination because it holds up.

Does the direct mark have to include both plain-text and a barcode?

Not necessarily. On the label, both a plain-text UDI and an AIDC barcode are required. For the direct mark itself, 21 CFR 801.45 is more flexible: it may be provided as easily readable plain text, as AIDC, or both — or through an alternative on-demand technology such as RFID. Available space on a small instrument is often the deciding factor. The non-negotiable is that whatever form is applied stays legible or scannable through the device's expected reprocessing life.

How is barcode quality graded and verified?

Marks are graded A, B, C, D, or F on a 4.0-to-0.0 scale. The applicable standard depends on the carrier: linear codes are graded per ISO/IEC 15416, printed 2D symbols per ISO/IEC 15415, and directly marked 2D codes per the AIM DPM guideline, ISO/IEC TR 29158 (elevated to the full standard ISO/IEC 29158:2020). For 2D symbols the overall grade is the lowest of the assessed parameters; for 1D linear symbols under 15416 the symbol grade is the average of the individual scan grades. Because 29158 accounts for the varied substrates and lighting of direct marks, a DPM symbol often grades differently than it would under 15415. Verification is not a one-time check — confirm the grade holds across reprocessing cycles, not just at first mark.

Selling into the EU: a separate UDI system

How is UDI marking for the EU different from the U.S.?

The EU runs a separate UDI system under the Medical Device Regulation (EU 2017/745), with the IVDR (EU 2017/746) covering in-vitro diagnostics. It uses the EUDAMED database and adds a Basic UDI-DI — a higher-level regulatory grouping identifier with no direct equivalent in the U.S. system. The concept mirrors FDA UDI, but the registry, identifiers, and timelines differ, so a U.S. GS1 GTIN and GUDID record do not automatically satisfy EU MDR. A device sold in both markets is managed against both systems, and the marking spec should reflect that from the start.

Keeping the marks, the data, and the roles aligned

What keeps the label, the barcode data, and the GUDID record in sync?

Consistency across three places — the human-readable label, the AIDC-encoded data, and the GUDID record — is where UDI programs quietly fail. The DI printed, the DI encoded in the barcode, and the DI on file in GUDID must match exactly, and the PI on the label must trace back to real production records. That alignment is a documentation and change-control problem as much as a printing one: when a device version changes, all three have to move together. A disciplined quality system is what prevents drift.

Can a contract manufacturer assign our UDI or submit to GUDID for us?

No — and the boundary is worth stating precisely. UDI assignment, the FDA-accredited issuing-agency account, labeling ownership, and GUDID submission are the labeler's legal responsibility, and the labeler is usually the OEM or specification owner. A contract manufacturer applies the marks and labels the OEM specifies, and maintains the lot, serial, and date traceability that feeds the UDI-PI, but it does not file with FDA or own UDI compliance. Getting this clear in the quality agreement up front prevents the most common program-level confusion.

Common UDI marking mistakes

Most UDI problems are designed in, not introduced at build — and several echo the root causes behind PCB and cable assembly failures. These patterns show up most often at inspection and in field scanning.

  • Barcodes that grade too low to scan. A symbol that looks fine to the eye can fall below a passing grade once aperture, contrast, and cell size are measured — verify to the right standard before production, not after.
  • Direct marks that fade or corrode after sterilization. A mark validated on a fresh part but not across reprocessing cycles will drift out of grade over its service life.
  • Mismatch between the label, the encoded data, and GUDID. The printed DI, the encoded DI, and the GUDID record must be identical; a version change that updates one but not the others breaks traceability.
  • Missing or incorrect human-readable text. A compliant barcode is not enough on the label — the plain-text UDI has to be present and correct.
  • Poor contrast or placement. Low mark-to-substrate contrast, or a code placed where handling and reprocessing wear it away, defeats an otherwise valid symbol.
  • A data carrier too dense for the surface. Forcing a high-density Data Matrix onto a surface the method cannot resolve guarantees scan failures — size the carrier to the part.

How i-TECH e-Services supports UDI marking and traceability

i-TECH e-Services is an ISO 13485:2016 contract electronics manufacturer that builds medical PCB assemblies and complete box-build systems for device OEMs. In UDI terms, i-TECH applies the UDI labels and both the human-readable and AIDC marks the OEM specifies during assembly and box build, and can coordinate direct part marking where 21 CFR 801.45 calls for a permanent mark. It maintains the lot, serial, and date traceability that feeds the UDI-PI, so each unit's encoded production data matches its physical mark and its build records.

The ISO 13485 quality system is what makes this dependable: it supplies the device history records, documentation, and change control the OEM folds into its own regulatory file, and in-process and final inspection can include scan-readability checks against the OEM's acceptance criteria so marks are confirmed scannable before parts ship. What stays with the OEM is equally clear — UDI assignment, the issuing-agency account, and GUDID submission remain the labeler's responsibility. i-TECH applies the marks and can confirm they scan; it does not file with FDA. Our medical device manufacturing page has the full picture, and our quality and testing capabilities cover the inspection side.

Planning the marking and traceability approach for a new medical build, or tightening one that already exists? Talk with i-TECH's team about how UDI labeling, direct-mark coordination, and lot and serial traceability fit into your assembly process.

Bottom line

UDI marking comes down to a few decisions made in the right order. Confirm the device is in scope and its class deadline has passed — all have, as of 2026; assign the UDI-DI through GS1, HIBCC, or ICCBBA and register it in GUDID; put both a human-readable UDI and an AIDC barcode on the label; and decide whether the device is reusable and reprocessed, because that, not preference, triggers a permanent direct mark under 21 CFR 801.45. Choose a data carrier and method sized to the part, verify to ISO/IEC 15416, 15415, or 29158, and confirm the grade holds through the device's life. Get the specification and the traceability right, and the marks take care of themselves.