Medical Device Manufacturer List Usa Data: Dataset Structure and Field Coverage

That filing obligation is what makes the list complete in a way no vendor directory can claim. Establishments must register with the FDA annually, so the file is a compliance artifact rather than a marketing opt-in: a contract sterilizer in Indiana and its OEM customer abroad appear side by side because both are required to file, not because either chose to be listed. Documented coverage runs from 2007 to present.

The registration{} block carries the account identity. You get the registration number, the FEI number that keys the whole file, a status code where 1 means Active, an expiry year that lapses each December 31, full address with ISO country code, dba-style name variants such as "DERMA SCIENCES CANADA, INC (dba DERMA SCIENCES, INC.)", an initial importer Y/N flag, a nested owner_operator{} tying the site to its parent firm, and a us_agent{} block naming the designated US contact for foreign registrants - which is frequently the exact person an outbound team wants.

What the facility does and sells sit in top-level arrays. establishment_type[] separates contract manufacturer, contract sterilizer, repackager/relabeler, reprocessor of single-use devices and foreign private label distributor, so you can segment outsourcers from owners of production. proprietary_name[] lists brand names and model numbers like "Burnfree 4 Oz and 2 Oz"; k_number and pma_number reference premarket clearances and approvals. The products[] array then lists each three-letter product code under 21 CFR Parts 862-892, resolved through an openfda{} join into device_name, device_class, regulation_number and medical_specialty_description such as Part 888 Orthopedic Devices.

Every pair carries a geocodable address, so territory carve-outs and radius-based account assignment fall out of the same load. One caveat FDA documents itself: products.created_date is flagged as possibly unreliable, so treat it as indicative when timing a prospect's listing history rather than as a definitive start date.

Two access routes serve the same record universe, and the choice is purely mechanical:

Field-level documentation publishes as PDF, XLSX and YAML reference files at if you need column definitions while loading.

The honest answer starts with the denominator. 333,804 records counts facility-product pairs across US and foreign establishments, not companies - one manufacturer listing five product codes contributes multiple rows. Deduplicating by FEI number collapses those to establishments, and grouping again by owner_operator{} collapses establishments to corporate families, which is the level most outbound teams actually target. Because registrations lapse annually, active establishments at any moment differ from the total-record count, so date-bound the status fields when you build.

Segmentation depth is where the file earns its keep over a purchased list. Every row carries both the facility class and the product classification, so the same load yields a contract-sterilizer universe, a Part 888 orthopedics universe, or a Class III specialist universe without buying anything extra. Add the related openFDA Device 510(k) Clearances API, shared from the health-care-supplies slice, and k_number and pma_number values on each listing resolve against 175,814 premarket notifications dating back to 1976 - applicant names there give you the brand-facing entity behind a contract site.

For demand context around each cluster, the CMS Data Hub - Medicare & Medicaid Datasets add utilization and reimbursement signals across 159 datasets, useful for prioritizing territories where procedure volumes support your price point.

A flat list of 333,804 pairs needs an ordering before anyone dials anything. Ranked by how quickly each screen converts raw registrations into qualified accounts:

1. Active-status establishments in your product codes. Filter products.product_code to your category and drop non-active rows first; this single cut typically removes the majority of noise in the file.
2. Establishment types matching your motion. establishment_type[] distinguishes contract manufacturer from repackager/relabeler and foreign private label distributor - pick the classes your product actually serves instead of calling them all.
3. Domestic sites for service-heavy deals. US-addressed establishments avoid the us_agent{} indirection and ship faster for field-service commitments.
4. Foreign firms with fresh listings. New entrants registering through a us_agent{} are actively investing in US market entry and often lack incumbent supplier relationships.
5. Owners with clearance activity. Cross-reference k_number and pma_number values against the monthly 510(k) file so firms clearing new devices get priority over dormant registrations.
6. Safety-signal exclusions. Screen the shortlist against the openFDA Device Adverse Events (MAUDE) API, whose 25,711,469 reports as of the 2026-08-11 data load reveal firms already under quality pressure - valuable if you sell remediation, disqualifying if you do not.

Plan around five documented constraints rather than discovering them mid-build:

None of these are blockers; they are reasons the file rewards a batch pipeline over naive paging. For how this dataset fits beside the other five pooled sources, see the health care equipment data guide.

Go deeper on the source itself at the openFDA Device Registration & Listing API dataset page, which documents every field, rate limit and verification result, or browse the best health-care-equipment datasets ranking and the health-care-equipment data hub indexing the whole slice.