Yes, you can plagiarize yourself in research, in the one sense that matters: reusing your own earlier text or findings in a way that lets readers believe they are new. Journals treat it as a named offense, publication ethics bodies have written guidelines for it, and graduate school is where most researchers meet it for the first time, usually by surprise.
The surprise is understandable, because the everyday logic of theft does not fit. You cannot steal your own sentences. What you can do is misrepresent the scholarly record, and that is the actual charge behind the awkward word. Here is where the lines sit, in publishing and in a graduate program, and how to reuse your own work without stepping over them.
What the reader is entitled to assume
A research text carries an implicit claim that a personal essay does not: this is new. New words, new analysis, and above all new results. Every rule about self-plagiarism unpacks from that one assumption.
The publishing world's preferred term is text recycling. The guidelines endorsed by the Committee on Publication Ethics define it as reusing your own previously published prose, a sentence, a paragraph, or a longer passage, in a new manuscript without quotation marks or a citation to the earlier publication. Notice that the definition is about concealment, not reuse. The same passage, disclosed, is a quotation. Undisclosed, it is recycling.
Reusing data is the more serious cousin. Publishing substantially the same results twice, without each version pointing at the other, is redundant publication, and journals treat it as a different and bigger problem than repeated prose, because it distorts the evidence base itself: one experiment starts counting as two.
The spectrum, from tolerated to sanctionable
Not all reuse is equal, and the guidance is more forgiving than most students fear on one specific point. Methods sections are where recycling is most common and most tolerated. There are only so many ways to describe the same protocol on the same equipment, and the COPE-endorsed guidance treats brief methods reuse as widely accepted, while the federal Office of Research Integrity allows boilerplate descriptions of standard methodologies and notes that reuse of unpublished internal documents, a grant application, an IRB protocol, is a different and safer category than reuse of published text.
The tolerance runs out fast outside methods. Recycling substantial parts of an introduction or discussion raises what the guidance calls more serious concerns, because those sections are exactly where the this-is-new claim lives. And ORI's writing guideline is blunt about the default: avoid reusing your own previously published text unless it is done in a manner that alerts readers.
At the far end sits salami slicing, cutting one study into several thin publications to multiply the line on the CV. That is not a text problem at all. It is a units-of-knowledge problem, and it is the version of self-reuse most likely to end in corrections or retraction.
The copyright twist nobody warns you about
Here is the part that makes self-plagiarism more than an etiquette rule. When your article was published, you likely signed an agreement transferring copyright, or an exclusive license, to the publisher. Your own paragraphs may no longer be yours to reproduce freely. ORI flags this directly: subsequent papers by the same author must be careful not to reproduce material that has previously been published.
So covert reuse of your own published text can be, simultaneously, an ethics problem with your new journal and a legal problem with your old one. The fix for both is the same disclosure, which is convenient, and one more reason the disclosure is not optional paperwork.
The grad school cases, one by one
Coursework into the thesis. Governed by your institution, not by journals, and usually by the same permission rule that covers submitting one paper to two courses. Ask your supervisor and check the program handbook before building a chapter on a seminar paper. The asking is the protection, as with every other form of accidental self-reuse.
Thesis chapters into journal articles. This is the normal life cycle of a dissertation, and nobody serious considers it self-plagiarism. Two conditions: tell the journal the manuscript draws on your thesis, and check the journal's policy on prior dissemination, since a public thesis repository counts as disseminated for some journals and not others.
Published articles into the thesis. Many programs explicitly allow a thesis by publication, with rules about framing text and co-author statements. The rules exist, which means the version without the rules followed is a finding waiting to happen. Get the format requirements in writing early.
Your prior paper as a source. Once something of yours is published, cite it the way you would cite a stranger, because to the reader that is what it is: a prior text in the record. Quote it with marks, paraphrase it with a citation, and remember that the reference list records what you actually read and used, your own work included.
Preprints and conference papers. Posting a preprint and then submitting to a journal is standard practice nearly everywhere now, but it works because it is visible: the preprint is linked, the journal knows, nothing is concealed. The same logic governs the conference-to-journal pipeline in fields that use it. An extended journal version of your conference paper is a recognized genre, and the recognized way to do it is to say so in the manuscript, cite the conference version, and check the venue's stated policy on how much overlap it accepts. The pattern across every legitimate case is identical: the earlier version is declared, not discovered.
How to reuse honestly, in one paragraph
Tell the editor at submission what overlaps and with what. Cite the earlier work where the overlap occurs, with quotation marks where the reuse is verbatim. Rewrite introductions and discussions fresh, recycle methods sparingly and with a pointer, and never let the same data appear in two places without each acknowledging the other. Disclosure converts nearly every version of this offense into ordinary scholarly practice, which tells you what the offense actually is.
If it goes wrong anyway
Discovered covert recycling in a manuscript usually ends at desk rejection and a note in the journal's file. In a published paper it means a correction or retraction, which is a public event attached to your name. Inside a graduate program it becomes an academic integrity case, and the process that follows is the standard one, with the extra sting that research misconduct findings travel with you in ways coursework findings do not.
A practical safety net before any submission: scan your manuscript and see what it matches, including your own earlier publications, which is the overlap you are least able to see because you wrote both sides of it. PlagiarismCheck.io is free, needs no account, and links each match to its source, so ten minutes tells you exactly which of your paragraphs are traveling with you. Then disclose them, and they stop being a problem.
The rule underneath all of it is short. Your reader assumes new. Anywhere that assumption would be wrong, say so, and self-plagiarism becomes impossible to commit.