UDI Barcode Generator

The FDA UDI rule: phased by device class, starting at class III

In the United States the FDA UDI rule places the obligation on the labeler — the entity that labels a device for commercial distribution, which is usually the manufacturer but can also be a specification developer, repackager or relabeler. The labeler has to do two things. First, put the UDI on the device label and on each package level in both plain-text and AIDC (machine-readable) form, with direct part marking additionally required for devices intended to be reused and reprocessed between patients. Second, submit and maintain the device identifier record in the GUDID, the FDA's Global Unique Device Identification Database, so that the DI printed on the package resolves to a published product record. Compliance was rolled out class by class, beginning with class III devices and working down through class II to class I and unclassified devices, with separate milestones for labeling, direct marking and data submission. If you are working out where your product sits, check the current FDA timelines for your device class rather than relying on a single cutover date.

EU MDR UDI: EUDAMED and the Basic UDI-DI

The EU Medical Device Regulation runs its own UDI system on the same GS1 building blocks, so the symbol this tool produces is structurally the same, but the registration side differs. Device data is registered in EUDAMED, the European database on medical devices, and EU MDR adds one concept that has no US equivalent: the Basic UDI-DI. That is a model-level identifier which groups devices sharing the same intended purpose, risk class and design characteristics; it never appears on the label, but it is the key that ties the declaration of conformity, certificates and technical documentation together in EUDAMED. What goes on the label is still the UDI-DI plus UDI-PI, exactly as generated here. As with the FDA, the application dates are staggered by risk class, so check current EU timelines for your device class.

Issuing agencies: GS1, HIBCC and ICCBBA

You cannot invent your own device identifier. It has to come from an issuing agency accredited by the relevant regulator, and the same handful of organisations are recognised on both sides of the Atlantic: GS1 (GTIN-based, by far the most widely used), HIBCC (the Health Industry Business Communications Council, whose HIBC standard is common among instrument and orthopaedic manufacturers), and ICCBBA (ISBT 128, used for blood, cell, tissue and organ products); the EU additionally designates IFA GmbH. Pick one system and stay inside it — do not mix encodings for the same product. This tool generates GS1 codes only: a GTIN-14 DI, GS1 application identifiers for the PI, and either GS1 DataMatrix or GS1-128 as the carrier. If your organisation issues under HIBCC or ICCBBA the data structure is different and you should follow that agency's specification instead.

Print requirements: getting it on the label is not the same as getting it read

A UDI carrier has to stay readable for the working life of the device, which is a harder target than an ordinary retail barcode. Four things decide whether it does. Contrast: print dark on light with enough ink density, and avoid reversed-out, coloured or coarsely screened backgrounds that blur the module edges. Size: keep the DataMatrix X-dimension at 0.254 mm or above; a cramped label is not a licence to shrink the symbol indefinitely. Quiet zone: at least one module clear on all four sides of a DataMatrix, at least 10 X either side of a GS1-128. Durability: devices that go through steam autoclaving, ethylene oxide or gamma irradiation need substrates and inks qualified for that process, otherwise the symbol yellows, blisters or the bars gain width after sterilisation and the grade collapses. Verify production samples to ISO/IEC 15415 before you commit to a run; grade C (1.5) is the commonly accepted floor.

The three mistakes that come up again and again

First, a wrong check digit. The last digit of a GTIN-14 is a Mod-10 check digit: weight the digits alternately by 3 and 1 from the right, sum them, and take the difference to the next multiple of 10. Get one digit wrong by hand and the symbol still prints perfectly — it just fails validation when someone scans it into a system. Second, AI order and separators. A variable-length AI such as (10) or (21) must be followed by an FNC1 separator if any further field comes after it; homemade element strings that bury a variable-length field in the middle without the separator produce data that lands in the wrong fields on scan. Third, date format. GS1 (11) and (17) are always six digits, YYMMDD; writing YYYYMMDD makes the AI the wrong length, and this tool will reject it rather than encode it.