FDA UDI Direct Marking: The Exception Needs Its Own File

Under 21 CFR 801.45, a device intended to be used more than once and reprocessed before each use must carry a permanent UDI mark on the device itself — not only on its label and in GUDID. The last compliance tier, for Class I and unclassified devices, closed in 2022. Direct marking is not a future obligation anyone is still ramping up to. What still catches device teams, years after the dates passed, is the exception: claiming one is not a labeling-team judgment call. It is an engineering determination FDA expects to find documented, not just asserted.

A separate requirement from the UDI label

Most device teams have UDI and GUDID compliance built into their labeling process: a UDI in plain text and AIDC form on the label, with the corresponding device record submitted to FDA's Global Unique Device Identification Database. Direct marking is a distinct, additional requirement layered on top of that baseline. It applies only when a device is intended to be used more than once and intended to be reprocessed before each use — both conditions, not either one. A single-patient reusable device that is never reprocessed between uses, or a device used once and discarded, does not trigger it on that basis alone.

Three exceptions, none of them automatic

FDA's direct-marking guidance and 21 CFR 801.45(d) set out exactly three grounds for skipping direct marking, and the agency has been explicit that invoking one is a documented decision, not a default. Teams that cleared a device through 510(k) or De Novo are expected to analyze, and where appropriate test, whether marking would compromise the device before concluding it would — and to keep that analysis in the design history file, where it will be the first thing an investigator asks to see if direct marking is missing.

1. Marking would interfere with safety or effectiveness. This requires an engineering analysis — and often testing — showing the mark itself (its process, location, or materials) would compromise the device, not a general sense that marking seems risky.
2. Marking is not technologically feasible. A narrow claim: the device's size, material, or surface genuinely cannot accept a durable, legible mark using any available method — not that the preferred method is inconvenient.
3. The device is single-use and is subjected to additional processing for one additional single use. This covers one further single use after reprocessing-for-disposal-avoidance scenarios — it does not extend to devices reprocessed for repeated, ongoing reuse.

An exception is an engineering determination FDA expects to find in the design history file — not an explanation you offer for the first time during an inspection. Why the documentation is the exception

Where this actually breaks

Two patterns account for most of the exposure we see. The first is scope confusion: a team treats a reusable Class II or Class III device as covered by the UDI label requirement and never separately scopes the direct-marking trigger, discovering the gap only during a device listing update or an audit of legacy instrumentation cleared years before the applicable compliance date. The second is an undocumented exception: engineering or regulatory concluded informally that marking "wouldn't work" on a given substrate, but no analysis was ever written down or filed. Both are correctable, but neither is a quick fix once FDA is already asking the question. For a sense of how a parallel UDI regime handles its own data-quality failure points, see how EU MDR UDI data quality breaks down on the Basic UDI-DI side — the mechanics differ, but the lesson that the obligation is not satisfied by the label alone is the same.

None of this is exotic once it is scoped correctly. It is a bounded exercise: confirm which devices trigger 801.45, verify durability against the labeled reprocessing method, and make sure every exception in use traces to a dated, documented determination rather than institutional memory. Companies that treat direct marking as solved because the UDI label and GUDID submission are in order are the ones most likely to find the gap during an inspection rather than before one.