What the research credit is, and why it exists

The research credit under Section 41 — still commonly called the R&D tax credit — is a federal income tax credit for increasing research activities. It is not a deduction. A deduction reduces taxable income; a credit reduces tax, dollar for dollar. A company with $1 million of qualified research expenses might see a federal credit in the neighborhood of $60,000 to $100,000, every year the spending continues, plus a state credit in most states that offer one.

It is also one of the most misunderstood provisions in the Code. The credit rewards incremental spending over a base amount, not total research spending; it applies to routine product and process development far short of laboratory science; and it interacts — but is not synonymous — with the Section 174 deduction rules that caused so much grief between 2022 and 2024.

A short history: 1981 to permanence

Congress enacted the credit in the Economic Recovery Tax Act of 1981 as a temporary, two-year measure meant to counter what policymakers saw as declining American industrial research. It then expired and was reinstated sixteen times over the following three decades — occasionally retroactively, occasionally with a gap. For most of its life, no CFO could budget for it with confidence.

Two structural milestones matter for anyone claiming the credit today:

• 1989–1990: Congress replaced the original rolling-average base with the fixed-base percentage mechanism that still governs the regular method, tying a company's base amount to its research intensity during 1984–1988.
• 2006: The alternative simplified credit (ASC) arrived, offering a computation that looks only at the prior three years of qualified spending — a practical necessity for companies without usable 1980s records.

The Protecting Americans from Tax Hikes (PATH) Act of 2015 finally made the credit permanent, and added the two features that opened it to early-stage companies: the payroll tax offset for qualified small businesses (now capped at $500,000 per year — see the payroll offset brief) and the ability to claim the credit against alternative minimum tax for eligible small businesses. Nothing in the One Big Beautiful Bill Act of 2025 changed Section 41 itself; the OBBBA's research provisions rewrote the deduction rules under new Section 174A, a related but distinct regime covered in the credit versus Section 174.

The economic rationale

The credit rests on a standard market-failure argument: firms cannot capture all of the returns from research, because knowledge spills over to competitors, suppliers, and future entrants. Left alone, the argument runs, firms will underinvest relative to the social optimum, and a subsidy tied to research spending narrows the gap.

That is also why the credit is incremental by design. Congress did not want to pay firms for research they would have done anyway; it wanted to subsidize the marginal dollar. Hence the base amount — a proxy for the taxpayer's expected level of research absent the credit — and a credit computed only on spending above it. Whether the incrementality machinery actually isolates marginal spending is a fair question; the fixed-base percentage in particular is an artifact of 1980s data. But the design intent explains most of the credit's odd geometry.

Who claims it

The statute contains no industry restriction. In practice the largest dollar volumes come from manufacturing, information, and professional/technical services, but the credit is claimed by software companies, aerospace and automotive suppliers, food and beverage producers, engineering and architecture firms, agricultural businesses, and contract manufacturers. What unites them is not white coats but a fact pattern: employees paid to figure out how to make something work when the answer was not known at the outset.

Three groups deserve particular mention:

• Startups. A pre-revenue company owes no income tax, so a credit against income tax is deferred value at best. Since 2016, qualified small businesses — under $5 million in gross receipts, with no gross receipts before the five-taxable-year window ending with the claim year — can instead apply the credit against payroll taxes, up to $500,000 a year.
• Pass-throughs. The credit flows through to partners and S corporation shareholders, where it is subject to general business credit limitations at the owner level.
• Government and commercial contractors. Eligibility turns on the funded research exclusion — who bears financial risk and who retains rights in the work. That analysis is developed in the funded research exclusion.

What counts: the two gates

Every credit claim passes through two gates, in order.

Gate one: qualified research. Under Section 41(d), an activity qualifies only if it satisfies all four parts of a statutory test — a permitted purpose (new or improved function, performance, reliability, or quality of a business component), elimination of technical uncertainty, a process of experimentation, and reliance on principles of hard science or engineering. Section 41(d)(4) then excludes certain activities outright: research after commercial production, adaptation, duplication, surveys and studies, most internal-use software (subject to a high-threshold-of-innovation test), foreign research, research in the social sciences, and funded research. The full treatment is in the four-part test explained.

Gate two: qualified research expenses. Even for qualifying activities, only four cost categories generate credit: taxable wages for qualified services, supplies used in the research, 65% of most contract research payments, and amounts paid for the rental or lease of computers (the modern home of cloud computing costs). Notably absent: overhead, rent, depreciation, and — a perennial surprise — the cost of equipment itself. The categories and their documentation demands are covered in qualified research expenses.

Computing the credit: a brief overview

Two methods exist, and the taxpayer chooses annually on Form 6765.

The regular method pays 20% of QREs above a base amount. The base amount is the taxpayer's fixed-base percentage — historically, the ratio of QREs to gross receipts during 1984–1988, with separate rules for companies formed later — multiplied by average annual gross receipts for the prior four years, with a floor of 50% of current-year QREs.

The alternative simplified credit pays 14% of QREs above 50% of the average QREs for the three preceding taxable years. A taxpayer with no QREs in any of those three years gets 6% of current-year QREs instead. No gross receipts, no 1980s archaeology.

Layered on both is Section 280C: because research expenses that generate the credit would otherwise be deducted too, the taxpayer must either reduce its deduction by the credit amount or elect a reduced credit — the gross credit multiplied by (1 − 21%) at the current corporate rate. Most taxpayers make the election for simplicity, particularly at the state level. Fully worked examples of both methods and the 280C math appear in how to calculate the R&D credit.

The practical yield of the two methods differs less than the headline rates suggest, because the regular method's 20% applies over a gross-receipts-driven base while the ASC's 14% applies over a spending-driven one. For a growing company, either can win; for a company with high historical research intensity relative to receipts, the regular method's base can swallow the credit entirely.

Common misconceptions

"We don't do R&D — we're not a lab." The statutory standard is technological uncertainty resolved through experimentation, evaluated at the level of the business component. A manufacturer iterating on tooling designs, a food company reformulating for shelf stability, a software team architecting around a scaling constraint — all routinely qualify. The research need not be novel to the world; it must be uncertain to the taxpayer.

"The credit pays 14% (or 20%) of our research spend." Both rates apply only to spending above a base. Effective yields for steady-state claimants typically run 5% to 10% of QREs after the 280C adjustment. A company whose research spending is flat against a high base may see very little.

"If it's deductible under 174A, it's creditable under 41." The deduction base is broader than the credit base. Section 174A reaches overhead, depreciation on research equipment, and foreign research; Section 41 reaches none of these. Every QRE should be a Section 174-type expenditure, but the reverse does not hold.

"Small claims don't get audited." Exam rates are not published by claim size, and the IRS has invested heavily in research credit enforcement — refund-claim specificity requirements, the redesigned Form 6765 with Section G reporting, and litigation on substantiation. The honest framing: the credit is legitimate and Congress wants it claimed, but it is a documentation-intensive position. See audit defense.

"Losses make the credit worthless." Unused credits carry back one year and forward twenty under Section 39, and qualified small businesses can monetize against payroll tax currently. A pre-profit company that documents its credits now banks them for the profitable years.

When the credit is not worth pursuing

Neutrality requires saying so: the credit is not for everyone.

• Fully funded contractors. A firm doing time-and-materials development work where the customer bears all risk and keeps all rights has no creditable research, however technical the work.
• Trivial qualified spend. The compliance cost — time tracking, project documentation, the Form 6765 build — has a floor. Below roughly the low tens of thousands of dollars of annual credit, many taxpayers reasonably pass.
• Flat spenders under the ASC. If QREs have been level for years, the ASC yields 14% of half your spend — real money, but the incremental design means declining research budgets can produce no credit at all.
• Taxpayers unwilling to document. A credit claimed on estimates reverse-engineered years later is an audit liability, not an asset. The case law on this point is unforgiving.

Where to go next

The rest of this cluster takes each piece in depth: the four-part test, qualified research expenses, the computation methods, and what happens on exam. For the deduction side of the ledger — Section 174A and the OBBBA transition rules — start with the credit versus Section 174. And for the conceptual distinction that underlies all of it, see credits versus deductions.