What research misconduct means in the federal rules

Three acts, three states of mind, one standard of proof

The federal definition sits in one sentence. Research misconduct means fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results. The same sentence adds that research misconduct does not include honest error or differences of opinion.

A finding is not made on that definition alone. Three things must all be true. There has to be a significant departure from accepted practices of the relevant research community. The misconduct has to have been committed intentionally, knowingly, or recklessly. And the allegation has to be proven by a preponderance of the evidence.

Each of those three is a separate hurdle, and each is defined elsewhere in the same rule rather than left to intuition. To act intentionally means to act with the aim of carrying out the act. To act knowingly means to act with awareness of the act. To act recklessly means to propose, perform, or review research, or report research results, with indifference to a known risk of fabrication, falsification, or plagiarism. Preponderance of the evidence means proof by evidence that, compared with evidence opposing it, leads to the conclusion that the fact at issue is more likely true than not.

Accepted practices of the relevant research community is also defined: practices established by the rule itself and by the funding components, as well as commonly accepted professional codes or norms within the overarching community of researchers and institutions that apply for and receive the relevant federal awards. The standard is the field's own, not a regulator's freestanding opinion of good science.

What each of the three words actually covers

Fabrication means making up data or results and recording or reporting them. That is the whole definition. It is about invention, not about error in collection or analysis.

Falsification means manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record. The reach of that definition is wider than it first appears. It covers the equipment and the process, not only the numbers. It also covers omission — leaving a result out can misrepresent the record as surely as changing one.

Plagiarism means the appropriation of another person's ideas, processes, results, or words, without giving appropriate credit. The rule then does something unusual: it says in its own text what plagiarism does and does not include. It includes unattributed verbatim or nearly verbatim copying of sentences and paragraphs from another's work that materially misleads the reader regarding the contributions of the author. It does not include the limited use of identical or nearly identical phrases that describe a commonly used methodology.

The record all three attach to is defined too. Research record means the record of data or results that embody the facts resulting from scientific inquiry, in physical or electronic form. The rule lists examples: research proposals, raw data, processed data, clinical research records, laboratory records, study records, laboratory notebooks, progress reports, manuscripts, abstracts, theses, records of oral presentations, online content, lab meeting reports, and journal articles. A conference poster and a lab notebook page sit inside the same category as a published paper.

What the definition deliberately leaves out

The exclusions are as load-bearing as the inclusions, and they are the part most often lost when the phrase is used loosely.

Honest error is out, by the text of the definition itself. So are differences of opinion. The evidentiary rule reinforces this: in deciding whether the burden of proof has been carried, the finder of fact is directed to give due consideration to admissible, credible evidence of honest error or difference of opinion presented by the respondent.

Self-plagiarism is out. So are authorship or credit disputes, including disputes among former collaborators who participated jointly in the development or conduct of a research project. The rule states plainly that self-plagiarism and authorship disputes do not meet the definition of research misconduct. Those are real disagreements with real consequences for careers, and the federal rule routes them somewhere other than a misconduct finding.

The rule also does not close off an institution's own standards. Institutions may have standards of conduct different from the standards for research misconduct under the rule, and a federal finding — or the absence of one — does not affect institutional findings or actions taken under an institution's own standards. A university can discipline conduct that the federal definition never reaches.

One thing the rule does treat as evidence rather than as a separate offence is the disappearance of records. A respondent's destruction of research records documenting the questioned research is evidence of research misconduct where it is established by a preponderance of the evidence that the respondent intentionally or knowingly destroyed records after being informed of the allegations. A refusal to produce records the respondent claims to possess is treated the same way.

The six-year clock, and the two things that stop it

The rule applies only to research misconduct occurring within six years of the date the department or an institution receives an allegation. That is a real limit, and it is the reason many old disputes go nowhere.

Two exceptions suspend it. The first is the subsequent use exception. It applies where the respondent continues or renews an incident of alleged misconduct that occurred before the six-year limit through the use of, republication of, or citation to the portions of the research record alleged to have been fabricated, falsified, or plagiarized, for the potential benefit of the respondent. The rule spells out where that use can appear: submitted or published manuscripts, submitted grant applications, progress reports, posters, presentations, or other research records.

The practical effect is worth stating plainly. A figure published nine years ago is outside the window. The same figure re-used in a poster last year is not, because the re-use is itself inside the window. Where the exception appears to apply and an institution concludes it does not, the institution must document that determination and retain the documentation.

The second exception covers public health. If the oversight office, or an institution after consulting that office, determines that the alleged misconduct, if it occurred, would possibly have a substantial adverse effect on the health or safety of the public, the six-year limit does not apply.

What the rule reaches, and what it has nothing to do with

This regime is tied to a funding stream. It applies to every institution that applies for or receives Public Health Service support for biomedical or behavioral research, research training, or activities related to that research or training. It covers applications and proposals for that support, the supported research and training itself, related activities such as the operation of tissue and data banks or the dissemination of research information, and research records produced during that work.

One clause widens it usefully: it reaches research proposed, performed, reviewed, or reported, and any research record generated from it, regardless of whether an application or proposal actually resulted in an award. An unsuccessful grant application is inside the rule.

What sits outside is equally definite. The rule does not supersede or establish an alternative to the regimes for fiscal improprieties, the ethical treatment of human or animal subjects, criminal matters, personnel actions against federal employees, or whistleblower and retaliation matters. It does not prohibit or limit how institutions handle allegations that fall outside its definition or that do not involve the relevant federal support.

For anyone reading a study cited on a product page, the boundary matters. A privately funded study with no federal support does not fall under this rule at all. That is not a statement that such a study is unreliable; it is a statement that this particular oversight machinery, with its defined terms, its institutional duties and its published outcomes, is not the machinery watching it. The definitions in this rule are still the clearest available vocabulary for describing what would and would not count as a serious research-integrity failure — which is why they are worth knowing even where the rule itself does not apply.

The rule exists because Congress required it. The statute establishing the Office of Research Integrity directs the Secretary to establish, by regulation, a definition of research misconduct, an administrative process each funded entity must have in place, a process for the office to respond to information about misconduct, and standards protecting people who make good-faith allegations from retaliation. The definitions above are the regulation that instruction produced.

Sources

Each document below is named as it names itself, with the date printed on that document rather than the day it was read.

1. 42 CFR 93.103 — Requirements for findings of research misconduct, read in full for the three cumulative elements: significant departure from accepted practices, an intentional, knowing or reckless state of mind, and proof by a preponderance of the evidence — Electronic Code of Federal Regulations, National Archives and Records Administration, September 2026
2. 42 CFR 93.234 — Research misconduct, the one-sentence definition covering fabrication, falsification and plagiarism in proposing, performing or reviewing research or reporting results, and its exclusion of honest error and differences of opinion — Electronic Code of Federal Regulations, National Archives and Records Administration, September 2026
3. 42 CFR 93.211, 93.212 and 93.227 — the definitions of fabrication, falsification and plagiarism, including the plagiarism definition's own exclusions for commonly used methodology phrases, self-plagiarism and authorship disputes — Electronic Code of Federal Regulations, National Archives and Records Administration, September 2026
4. 42 CFR 93.104 and 93.105 — the six-year limitation with its subsequent-use and public-health exceptions, and the evidentiary standards including the burden of proof and the treatment of destroyed or withheld research records — Electronic Code of Federal Regulations, National Archives and Records Administration, September 2026
5. 42 CFR 93.102 — Applicability, read in full for the scope of the rule, its reach to unfunded applications, and the list of regimes it does not supersede — Electronic Code of Federal Regulations, National Archives and Records Administration, September 2026
6. 42 U.S.C. 289b — Office of Research Integrity, read in full for the statutory establishment of the office and the direction that the Secretary establish by regulation the definition of research misconduct, the required administrative processes, and standards preventing retaliation — Office of the Law Revision Counsel, United States House of Representatives, September 2026