The free tax research stack: primary authority without a subscription

Every layer of primary federal tax authority is available free, directly from the government that produced it. The Internal Revenue Code lives at uscode.house.gov, Treasury regulations at eCFR Title 26, revenue rulings and procedures in the Internal Revenue Bulletin, Tax Court opinions at ustaxcourt.gov, and legislative history at govinfo.gov. What paid services sell is convenience — editorial summaries, annotations, citators — not the law.

That distinction matters for anyone signing returns or defending positions. A research file built on primary sources is stronger than one built on a vendor's paraphrase, and the free stack now covers every tier of the authority hierarchy described in our guide to weighing tax authority.

The stack, source by source

Each tier of authority has one canonical free home; learn these seven and you rarely need anything else.

Two habits make the stack work. First, always start at the statute — the technique in How to read the Internal Revenue Code applies at every layer. Second, record where and when you found each item; free sources are current, but your memo needs to prove it was current when you relied on it.

Layer 1: the Code at uscode.house.gov

The Office of the Law Revision Counsel publishes the official United States Code, and its "preliminary" edition of Title 26 is updated continuously as acts are enacted. This is the fastest reliable way to read post-OBBBA statutory text — new Section 174A, the amended Section 41 coordination rules — without waiting for commercial compilations to catch up.

Three features earn their keep:

• Editorial notes. Below each section, amendment notes list every act that touched the text, with effective dates. When a client's issue spans 2022–2025, these notes tell you which version of the statute governs which year.
• Prior-text reconstruction. The amendment notes quote deleted language, so you can reconstruct, say, pre-OBBBA Section 174 for an amended-return year.
• Cross-reference stability. Statutory cites in your memo should point here, not to a mirror site of uncertain vintage.

Layer 2: regulations at eCFR and the Federal Register

The eCFR is the continuously updated Code of Federal Regulations — Title 26 holds the Treasury regulations. It shows the current text with source notes (the Treasury Decision number and Federal Register cite for each amendment). For anything time-sensitive, the eCFR's point-in-time feature lets you display the regulation as it stood on a chosen date.

The Federal Register is where regulations are born: proposed rules, final rules, and — critically — their preambles, where Treasury explains its reasoning, responds to comments, and states applicability dates. Preambles are underused; they deserve their own discussion, which we give them in Why regulation preambles matter.

Layer 3: the Internal Revenue Bulletin

Everything the IRS publishes with precedential intent — revenue rulings, revenue procedures, notices, announcements, final regulation reprints — appears in the weekly Internal Revenue Bulletin. The IRB archive on irs.gov reaches back decades. If a ruling is not in the IRB, it is not the IRS's official position; that line separates published guidance from private letter rulings and internal memoranda, a distinction unpacked in PLRs, TAMs, and CCMs explained.

Searching the IRB is the stack's weakest link — there is no elegant full-text search. The practical workaround: use a general search engine restricted to site:irs.gov/irb, or search the IRS's guidance pages for the topic and follow the IRB cites they give.

Understand the categories before weighing them. Revenue rulings apply the law to stated facts; revenue procedures prescribe compliance mechanics; notices often preview regulations and can be relied on in the interim; announcements are administrative housekeeping. Each carries a running status line — later items can modify, supersede, revoke, or obsolete earlier ones — so no IRB cite goes into a memo without a status check. The distinctions and their precedential consequences are unpacked in Revenue rulings vs. revenue procedures.

A separate free trove sits outside the IRB: the IRS's released written determinations — private letter rulings, technical advice memoranda, Chief Counsel advice. These bind no one but the taxpayers who got them, yet they show how the National Office actually reasons; treat them as intelligence, not authority.

Layer 4: cases at ustaxcourt.gov and govinfo

The Tax Court's DAWSON system at ustaxcourt.gov offers free full-text search of opinions — division (TC) opinions, Memorandum opinions, and Summary opinions — by name, judge, date, or keyword. Since most research-credit and cost-segregation disputes are litigated there, DAWSON alone covers a large share of specialty-tax case law.

Appellate and district court opinions live in govinfo's United States Courts Opinions collection, though coverage of older decisions is uneven; for pre-2004 circuit opinions you may need a law-library visit or Google Scholar's case-law search as a supplement.

Layer 5: legislative history at govinfo, JCT, and CRS

Govinfo hosts the raw materials: enrolled bills, public laws, Statutes at Large, committee reports, and the Congressional Record. The Joint Committee on Taxation publishes bluebooks (post-enactment general explanations), revenue estimates, and pamphlets prepared for markups. CRS reports — the Congressional Research Service's analyses, public since 2018 — are the best free plain-English overviews of tax provisions, with citations you can chase.

When to descend to this layer, and how much weight it carries in a textualist era, is the subject of Legislative history research.

A worked example: contract research under Section 41

Suppose a CPA asks: client funds research performed by a contract engineering firm; the contract pays fixed fees per milestone, and the client keeps the IP. Can the client claim 65% of the payments as qualified research expenses — and does it matter who bears the risk?

Step 1 — statute. At uscode.house.gov, read Section 41(b)(3): "contract research expenses" are 65 percent of amounts paid to any person (other than an employee) for qualified research. The statute is silent on funding risk from the payor's side — but Section 41(d)(4)(H) excludes research "funded by any grant, contract, or otherwise," which governs the researcher's side. Two provisions, one economic arrangement.

Step 2 — regulations. At eCFR Title 26, Treas. Reg. §1.41-2(e) defines contract research expenses, requiring the agreement be entered before the research is performed and that payment be for research rather than its product. Treas. Reg. §1.41-4A(d) supplies the funding test: amounts are not funded if payment is contingent on success (the researcher bears risk) — and rights to the research results matter too.

Step 3 — IRB guidance. A site-restricted search of irs.gov/irb for funded-research guidance turns up little published ruling authority — itself a finding worth recording, because it means the action is in regulations and case law.

Step 4 — cases. In DAWSON and govinfo, the funded-research line of cases: Fairchild Industries (Federal Circuit, risk follows payment contingency), Dynetics (fixed-price versus cost-plus terms parsed contract by contract), and Populous Holdings (Tax Court order finding fixed-price design contracts unfunded where the researcher bore correction costs and the client's rights were non-exclusive). Our briefs on Fairchild Industries and Dynetics walk the reasoning.

Step 5 — conclusion, with hierarchy. Fixed-fee milestone payments with client-retained IP point toward the client claiming 65% under §41(b)(3): the contractor is paid regardless of research success in each milestone only if the milestones themselves guarantee payment — the contract terms decide. The memo cites statute, regulation, and cases in that order, notes the absence of published rulings, and flags the risk that both parties claim the same research.

Time spent: roughly two hours, cost: zero. The same question through a paid service is faster to survey but ends in the same primary sources.

Statute (§41(b), (d))2 citations

Regulations (§1.41-2, -4A)3 citations

Published rulings0 citations

Cases3 citations

Legislative history1 citations

Authorities cited in the example memo, by tier; illustrative of a typical specialty-tax research file.

What the free stack does not do

Neutrality requires the caveats. The free stack has no citator — nothing flags that a case was reversed or a ruling obsoleted; you must trace subsequent history manually, a discipline covered in Citators and checking currency without paid tools. It has no editorial layer connecting authorities to each other; you build those connections yourself. Full-text search quality varies from excellent (DAWSON, Federal Register) to poor (IRB). And state materials are a separate, messier world.

For a firm doing daily tax research, a paid service earns its subscription in speed. But the free stack is the ground truth the paid services summarize — and for verifying the authorities that go into a signed return position, there is no substitute for reading the primary source.

There is also a coverage asymmetry worth naming. The free stack is strongest for current law and post-2000 materials; it thins for older items — pre-1995 Federal Register preambles require govinfo's scanned archive, pre-DAWSON Tax Court Memorandum opinions can be hard to locate, and Cumulative Bulletin volumes before the digital IRB era are scattered. For historical research on long-lived provisions (the Section 41 base-period rules, for instance, still turn on 1980s law), budget extra time or a library trip.

Building the habit

Bookmark the seven sources. Start every question at the statute. Save PDFs of everything you rely on, dated, into the client file — the memo format in Building a research file turns this raw material into penalty protection.

Two closing disciplines keep the stack honest. First, separate finding law from weighing it: the sources above answer "what exists," while the substantial-authority framework answers "what is it worth" — and a memo that skips the second step is a bibliography, not an opinion. Second, note what you did not find. The absence of published guidance on a question is itself a research result, one that changes how conservatively the position should be framed and whether disclosure is warranted. The stack is free; the discipline is the investment.