Most authorship disputes are not fought over who did the least work — they are fought over who remembers the work differently, months or years after the fact, once a manuscript is close to submission and the stakes (a publication line, a grant renewal, a dissertation chapter) are suddenly real. An authorship agreement is the tool research teams already have available to prevent that: a short, written record of who is expected to be an author, what each person’s contribution role is, and how order and disputes will be handled — signed before drafting starts, not negotiated for the first time when the manuscript is nearly done.
This guide covers what an authorship agreement is, why timing is the variable that determines whether it actually works, what a usable one covers, real examples of the author-attestation forms journals already require at submission (which show what a rigorous agreement looks like in practice), and a practical process for a team to adopt one.
What an authorship agreement is
An authorship agreement is a written understanding, reached by the members of a research team, that documents: who is expected to qualify as an author on the eventual output(s) of a project, what each person’s anticipated contribution is, how author order will be determined, and what happens if contributions or circumstances change. It is distinct from two things it is often confused with:
- It is not a legal contract in the contract-law sense for most teams — it is a documented understanding, not something typically drafted by counsel or filed with an institution’s research office, though large multi-site collaborations sometimes formalize authorship terms within a broader research collaboration agreement (RCA), which governs the institutional relationship rather than individual credit.
- It is not the same document as a CRediT contribution statement, which is published alongside the final paper and records what happened. An authorship agreement is made earlier and is a working plan for what is expected to happen — it gets revised as the project evolves, and it is normally internal to the team rather than published.
The two documents are related, though: a team that sets contribution expectations early, in CRediT or ICMJE-criterion language, has an easier time producing an accurate published contribution statement later, because the categories were already being tracked from the start rather than reconstructed from memory at submission.
Why timing — before writing begins — is the part that matters
The advice to formalize authorship isn’t new; ICMJE, COPE, and most journal author guides all recommend authors resolve authorship “early.” What actually determines whether that advice is followed is a specific, narrow window: before the first full draft exists. A few reasons the pre-writing point is the one that works, and why waiting rarely does:
- Contribution categories are still forward-looking, not contested. Before a manuscript exists, “who will lead the analysis” or “who is drafting the introduction” is a plan, not a claim about what already happened. Once a draft exists, the same conversation becomes an argument about whether someone’s actual contribution was big enough to justify authorship or a particular position in the order — a much harder conversation, and one that now has an emotional stake attached to it.
- It’s easier to add or drop a contributor cleanly before writing starts. A collaborator who turns out not to be needed, or a new contributor brought in partway through, is a simple update to a plan made in advance. The same change made after a draft circulates looks like someone being pushed out of, or squeezed into, authorship — which is exactly the pattern behind many authorship disputes and a recurring driver of gift and honorary authorship problems.
- It gives the team a shared reference point instead of relying on memory. Multi-year projects, trainee turnover, and multi-site collaborations all make “who agreed to what” hard to reconstruct accurately a year or two later. A dated, written agreement removes the need to rely on anyone’s recollection.
- It surfaces disagreement while it’s still cheap to resolve. If two people both expect to be first author, that’s a solvable problem in month one. It is a much more damaging problem in the week before submission, when the manuscript itself becomes a bargaining chip.
None of this requires certainty — contribution plans change as projects evolve, and a good authorship agreement is treated as a living document, revisited at defined checkpoints (a new dataset, a new site joining a study, a change in project direction), not a one-time form signed and forgotten. The point of doing it early is establishing the default and the process for updating it, not locking in a final, unchangeable author list on day one.
What an authorship agreement should cover
There’s no single mandated format — unlike, say, a data management plan, no funder or standards body requires a specific authorship-agreement template. But the agreements that hold up, and the attestation forms journals require at the other end of the process (see the real examples below), consistently cover the same ground:
1. Who is expected to be an author, against a real threshold
The team should agree on the standard it is using to decide who qualifies as an author at all — not just who is “involved” in the project. The most widely used standard is the International Committee of Medical Journal Editors (ICMJE) four-criteria test, which requires all four of the following, not any one:
- Substantial contributions to the conception or design of the work, or the acquisition, analysis, or interpretation of data;
- Drafting the work or reviewing it critically for important intellectual content;
- Final approval of the version to be published; and
- Agreement to be accountable for all aspects of the work, ensuring questions about its accuracy or integrity are appropriately investigated and resolved.
ICMJE is explicit that anyone contributing meaningfully but not meeting all four criteria should be acknowledged, not listed as an author — and that authorship qualification is the author group’s own responsibility, not the journal’s, to work out. An authorship agreement is essentially the mechanism by which a team applies that standard to itself, in advance, rather than improvising it at submission. See CASRAI’s CRediT vs. ICMJE comparison for how the ICMJE criteria (a qualification threshold) relate to CRediT (a role taxonomy for describing what qualifying authors actually did).
2. Contribution roles, ideally in a standard vocabulary
Once the team knows who is expected to qualify, recording each person’s anticipated role in a consistent vocabulary — the 14 CRediT roles (conceptualization, methodology, data curation, formal analysis, writing – original draft, writing – review & editing, supervision, funding acquisition, and so on) are the most widely adopted option — avoids vague, disputable language like “helped with the study” later. See how CRediT roles can be held by more than one author, and how a role can be shared, since most real projects don’t map one role to one person cleanly.
3. Author order, and how it will be decided
Order conventions vary substantially by field — CASRAI’s guide on how authorship order is decided across disciplines covers first-author, last-author/senior-author, and alphabetical conventions in detail. An authorship agreement should state which convention the team is using and, where order depends on relative contribution rather than a fixed convention (e.g., “first author goes to whoever leads the analysis”), what that determination will be based on.
4. What happens if contributions change
Because the agreement is made before contributions are final, it should say what happens when reality diverges from the plan — someone contributes less than expected, someone new joins partway through, or someone’s role shifts substantially. A brief, agreed process for revisiting the agreement at defined points is more useful here than trying to anticipate every scenario in advance.
5. A named point of escalation for disagreement
Most disagreements are resolved by the author group itself. Where they aren’t, the agreement should say what happens next — typically escalation to the PI or a neutral third party, and ultimately the institution, which is the standard jurisdictional order under ICMJE and COPE guidance (see how authorship disputes actually get resolved through the institutional process and CASRAI’s guide to resolving authorship order disputes specifically).
Real examples: what journals already require at the other end of the process
Research teams don’t have to invent this from scratch — most journals already require something functionally similar at submission, which is useful both as a model and as a reason to start the conversation earlier rather than leave it until this form lands in the corresponding author’s inbox.
The Journal of Pediatrics requires a signed “Authorship Agreement and Contribution” form from every author before publication. The form requires each author to confirm they meet all four ICMJE conditions, to check off which specific contributions they made from a checklist based on CRediT-style categories (conceptualization/design, methodology, investigation, supervision/oversight, funding acquisition, data curation, formal analysis, resources), to indicate participation in drafting and/or review of the manuscript, and to separately attest “yes” or “no” to giving final approval and agreeing to be accountable for the work. It also requires disclosure of any prior publications with overlapping data and confirmation the manuscript isn’t under consideration elsewhere. It is, in effect, exactly the kind of document a research team should be filling in collaboratively from the start of a project, not reconstructing under deadline pressure at submission.
AHA/ASA journals (the American Heart Association’s journal family, including Circulation) similarly require authors to complete an authorship attestation confirming they meet ICMJE’s criteria and identifying their specific contributions to the work, in addition to a separate conflict-of-interest disclosure and copyright/license agreement, per the journals’ published submission guidelines.
The pattern across both is the same: a signed record, tied to specific ICMJE criteria and specific contribution categories, confirming who did what and that everyone agrees to be accountable. Building that record early, as a team agreement, rather than treating it purely as a submission-time compliance step, is what this guide is arguing for — the form doesn’t have to originate at the journal.
How disputes get handled when a prior agreement exists
An authorship agreement doesn’t guarantee a dispute-free project, but it changes what a dispute looks like when one happens: instead of two people with conflicting, unrecorded expectations, there’s a dated document both sides can refer back to, and — if the agreement included a revisit process — a record of when and why expectations changed. That materially shortens the institutional process described in CASRAI’s guide on how authorship disputes actually get resolved, and it helps distinguish an ordinary disagreement over credit from something more serious — see when an authorship dispute crosses the line into research misconduct and COPE’s own guidance, summarized in CASRAI’s walkthrough of COPE’s flowchart for suspected ghost, guest, or gift authorship.
A practical process for adopting one
- Raise it at project kickoff, not at manuscript planning. The natural moment is when a study, grant, or collaboration is first scoped — before any writing, and often before data collection is complete.
- Use the ICMJE four-criteria test as the qualification threshold, and record each person’s anticipated contribution against it, ideally in CRediT-role language so it maps cleanly onto a published contribution statement later.
- State the author-order convention being used, and who makes the final call if it isn’t self-evident from contributions alone.
- Set revisit points — a natural point is whenever the scope of the project materially changes (new data, new site, new significant contributor) — rather than treating the initial agreement as final.
- Name an escalation path for disagreements the group can’t resolve on its own, consistent with the institutional process most journals and standards bodies already expect (author group first, then a neutral party, then the institution).
- Keep it dated and accessible to every listed contributor — a shared document the whole team can see, not something held privately by the PI.
None of this requires new infrastructure. A short document, even a shared spreadsheet or a paragraph in a project’s shared drive, that names contributors, roles, and order expectations, dated at the start of a project, does the job. The value isn’t in the format — it’s in doing it before the writing starts, while the conversation is still about a plan rather than a grievance.
