The R&D Tax Credit · Brief · Working level
Medical device companies: 510(k), PMA, and the research credit
How device development maps to Section 41 across the 510(k) and PMA pathways — prototyping, verification and validation testing, and where regulatory submission work stops qualifying.
Medical device development is regulated experimentation: FDA design controls under 21 CFR 820.30 force manufacturers to document design inputs, outputs, verification, validation, and changes — very nearly the record Section 41 asks for. The credit analysis therefore tends to be about boundaries rather than substance: where the qualifying engineering ends and excluded regulatory administration, routine testing, and production begin. The regulatory pathway — 510(k) versus PMA — changes the shape of the spend, not the legal test.
Pathway does not decide qualification — uncertainty does
A common misconception is that 510(k) devices cannot qualify because the pathway rests on substantial equivalence to a predicate. Regulatory equivalence is not engineering certainty: a company can know its device is "equivalent" in intended use and still face open questions about materials, mechanism, miniaturization, firmware, and manufacturability that only iterative design and testing resolve. The four-part test asks whether the appropriate design was uncertain at the outset and whether alternatives were systematically evaluated — questions answered by the design history file, not the FDA docket. PMA-class devices (novel, higher-risk, requiring clinical evidence) simply carry more of the same, plus clinical study costs that follow the analysis in the biotech and pharma guide, including the foreign-site exclusion and CRO rights-and-risk review.
Where the credit lives across a device program:
| Program stage | Typically qualifies? | Notes |
|---|---|---|
| Concept and feasibility engineering | Yes | Alternatives evaluation; capability uncertainty |
| Prototype design and build iterations | Yes | Wages plus materials consumed as supply QREs |
| Design verification (bench, biocompatibility, electrical safety, human factors) | Yes, while design is fluid | Test failures feeding design changes are the proof |
| Design validation (does the device meet user needs) | Yes, to design freeze | Including simulated-use and cadaver labs |
| Process development, pilot line, test-method validation | Yes | Process business component, as in manufacturing |
| 510(k)/PMA assembly, predicate analysis, FDA correspondence | No | Administrative, not experimentation |
| Repetitive testing of the frozen design for submission record | No | Routine testing exclusion, §41(d)(4)(D) |
| Production lot-release and incoming QC testing | No | Quality control, excluded |
| Post-market surveillance and complaint handling | No | Post-commercial-production |
Prototypes and V&V: the two big expense pools
Prototyping. Wages of design, firmware, and test engineers are the largest QRE category; materials consumed in bench units, tooling trials, and destructively tested builds are supplies under Section 41(b)(2)(C). The recurring adjustments mirror manufacturing: capitalized molds and test equipment are depreciable property, not supplies, and pilot-build units that ship as sellable inventory are production. Keep a disposition log — scrapped, tested-to-destruction, retained as reference, sold — for every build; the QRE guide covers the categories in detail.
Verification and validation. V&V is where device claims are most often over- and under-claimed at once. The qualifying core is testing that informs design — a failed fatigue test that sends the team back to geometry, usability findings that change the interface. Under Treas. Reg. §1.41-4, evaluating alternatives through modeling and testing is the process of experimentation itself, so pre-freeze V&V is not merely "support" but the experiment. After design freeze, reruns of the same protocols to generate a clean submission record are routine data collection; claim the engineering, not the paperwork run.
Device software — embedded firmware, companion apps, and increasingly ML-based functions — generally qualifies as software developed for sale as part of the product, avoiding the internal-use software hurdles. Claims are reported at the business-component level on Form 6765 Section G; the IRS research credit overview frames the general rules. The design history file the FDA already requires is the audit file the IRS wants — few industries get their substantiation this cheaply.
Frequently asked questions
- Does medical device development under a 510(k) qualify for the R&D tax credit?
- The engineering usually does; the filing does not. A 510(k) claims substantial equivalence to a predicate device, but getting there — design iterations, bench and biocompatibility testing, human factors work — routinely involves genuine design uncertainty that satisfies Section 41(d). What does not qualify is assembling the submission itself, predicate searches, and administrative FDA correspondence.
- Are verification and validation (V&V) testing costs qualified research expenses?
- Design verification and validation performed while the design can still change — testing that feeds failures back into design iterations — is part of the process of experimentation, and its wages and consumed test units are QREs. Repetitive testing of a frozen design for the record, and routine lot-release or quality-control testing in production, are excluded as routine testing under Section 41(d)(4)(D).
- Can prototype device units be supply QREs if they are later used clinically or sold?
- Prototype materials consumed in development — bench units, units destroyed in testing, early builds evaluated against a test plan — are supply QREs under Section 41(b)(2)(C). Units from pilot builds that are sold, or clinical units whose costs are reimbursed by a sponsor or payer, draw the production-cost and funded-research arguments; disposition records for every build are what keep the supply claim intact.