Responding to peer review when you disagree

A response to peer review needs to let an editor see what was requested, what the authors did and why any request remains unmet. Agreement is not required on every point. A traceable answer is.

Begin by reading the editor’s decision alongside all the reviews. The editor may prioritise particular issues, reconcile reviewers or impose requirements that are not repeated in the reports. Treat the decision letter as part of the revision task, not merely the envelope containing the reviews.

This guide provides a revision matrix and an annotated letter structure. All examples are blank templates. They do not invent study results, reviewer quotations or successful editorial outcomes.

Turn the reports into a work plan

Save an untouched copy of the decision and reviews. Create a working document with a stable identifier for each comment, such as Editor 1, Reviewer 1.1 and Reviewer 2.1. Keep the full comment available even when you split it into several tasks.

Separate requests for clarification, analysis, additional evidence, interpretation, references, presentation and compliance. One paragraph may contain several requests. Answering its first sentence does not address the rest.

Assign an owner and a second reader to consequential changes. The owner performs the work; the second reader checks that the manuscript and response accurately describe it. Do not let the corresponding author become the only person who understands the revision simply because they will upload the files.

William Stafford Noble’s guidance emphasises a clear, self-contained response that makes the changes easy to identify. Use that principle to organise the record, rather than copying a formulaic letter or treating politeness as a substitute for a substantive answer. 1

The revision matrix

Keep this matrix internal while planning. The final response can be more readable prose, but every request should have an accountable disposition.

Comment	Underlying concern	Action or reason not to act	Owner	Manuscript location	Status
Editor [number]	___	___	___	___	Open / checked
Reviewer 1.[number]	___	___	___	___	Open / checked
Reviewer 2.[number]	___	___	___	___	Open / checked
Cross-cutting consistency check	___	___	___	___	Open / checked

Use the concern column to identify what the reviewer needs to evaluate. A request for another figure may really concern an unclear comparison. A request for a new analysis may concern whether the principal conclusion is robust. Understanding that concern helps you decide whether the requested action, an alternative or a narrower claim is appropriate.

Do not silently replace the reviewer’s request with an easier task. In the response, acknowledge the actual request and explain how your chosen action addresses it. If it does not fully address it, say what remains unresolved.

Mark an item checked only when the change is present in the actual revision file, not when someone says they intend to make it. This simple distinction prevents promises in a response letter from being mistaken for completed work.

Answer first, then explain

For each point, begin with the disposition: changed, clarified, newly analysed, partly addressed or not changed. Follow with the substantive explanation and the location of the change. The reviewer should not have to read several paragraphs before learning what happened.

When you agree, avoid a reply that only says “Done”. Describe the modification sufficiently for the editor to assess it without searching the whole manuscript. Quote a short revised passage if that is the clearest way to show the result, then give a reliable location.

When you disagree, state the reason in terms of the research question, design, evidence or journal requirement. “We disagree” is a position, not an argument. “Beyond scope” also needs explanation: which question would the request answer, and why is that question distinct from the claim the paper makes?

Do not bury a new limitation in the response alone. If readers need it to interpret the study, put it in the manuscript as well. The response is a record of the editorial exchange; it should not become the only place where essential methods or qualifications are available.

Distinguish three forms of disagreement

The request rests on a misunderstanding. Identify the relevant evidence and consider whether the manuscript made it clear enough. You can explain that the procedure was already performed while still improving the description. Avoid using the misunderstanding as a reason to make no reader-facing change.

The request is feasible but scientifically inappropriate. Explain the assumptions it would require or the question it would answer. Consult the team member with the necessary expertise. If an alternative analysis addresses the concern more appropriately, describe it and report the actual result only after completing and checking it.

The request cannot be fulfilled with the available study. Explain the missing information or design limitation accurately. Lack of resources alone does not make a request scientifically unnecessary. The editor may reasonably conclude that the unsupported claim must be narrowed or that the paper cannot proceed in its current form.

These distinctions help avoid two unhelpful responses: agreeing to everything regardless of scientific sense, and treating every difficult request as an intrusion. The aim is a defensible account of the work’s limits and what the revision can establish.

When reviewers ask for incompatible changes

Identify the precise conflict. One reviewer may request extensive context while another asks for a shorter introduction. Those comments may be reconcilable through editing. Two reviewers may instead propose analyses that answer different questions or rely on incompatible assumptions.

Look for direction in the editor’s letter. If the priority remains unclear and the decision materially affects the revision, ask the editor a focused question before investing in extensive work. Include the comment identifiers, the conflict and your proposed resolution.

Use a short clarification template:

Reviewer [identifier] requests [action], while Reviewer [identifier] asks for [conflicting action]. We propose [resolution] because [reason connected to the manuscript’s question]. Would this address the editorial priority for the revision, or would you prefer [specific alternative]?

Do not ask the editor to settle an internal preference the team can resolve itself. Use the inquiry for consequential ambiguities, not every stylistic choice. While waiting, continue independent revisions that do not depend on the answer.

When submitting, explain the resolution to both reviewers. Do not answer each as though the other request did not exist, and do not claim the editor endorsed a choice unless the correspondence supports that statement.

Keep new analyses distinguishable from the original plan

If review leads to additional analyses, record what was requested, what was done and how it relates to the original design or analysis plan. Do not rewrite the history so that a post-review exploration appears to have been prespecified.

Check all dependent text and numbers after an analysis change: abstract, tables, figures, discussion, supplement and data or code records. A response can be technically correct while the manuscript still contains an older result elsewhere.

For health research, an applicable reporting guideline can help check that the revised account remains complete. EQUATOR recommends locating reporting items in descriptive sections, paragraphs, tables or supplements and explaining omitted items. The guideline remains a reporting aid, not proof that the analysis itself is valid. 3

Where a review request involves data sharing, compare it with consent, access permissions and the journal’s policy. PLOS’s data guidance recognises that legal or ethical restrictions may require a specified access arrangement. Do not promise public release of restricted material to make a response appear cooperative. 5

An annotated response-letter structure

Adapt this structure to the journal’s instructions. Keep the bracketed fields until you can replace them with checked facts.

Dear [editor],

Thank you for considering the revision of [title and manuscript identifier]. We have addressed the decision letter dated [date]. The principal changes are [two or three concrete completed changes]. The main unresolved issue is [issue, if any], addressed below at [comment identifier].

Editor comment [identifier]

[Insert the complete relevant comment from the private decision letter.]

Response: [Direct answer.] We changed [specific content] because [scientific or explanatory reason]. The revised material appears in [section, paragraph, table or figure; line numbers where useful].

Reviewer comment [identifier]

[Insert the complete relevant comment.]

Response: We have not made [requested change] because [specific reason]. To address the underlying concern, we have [actual change, alternative or explicit limitation]. This affects our interpretation as follows: [accurate consequence]. The manuscript now states [short revised text or location].

Files supplied: [Clean manuscript, marked manuscript, response and updated supporting files required by this journal.]

The opening identifies the revision rather than praising it. The direct answer makes the disposition visible. The location lets the editor inspect it. The disagreement paragraph acknowledges that an alternative may leave a limitation; it does not pretend that an explanation is equivalent to new evidence.

Do not retain a template sentence describing work you did not do. In particular, “we have verified all analyses” and “all authors approved” are factual statements that require actual verification and approval.

Handle reference requests on relevance

Read a suggested source before deciding whether to cite it. Check whether it supports the proposition for which the reviewer recommends it, whether the relevant version is current and whether it adds something the manuscript needs.

If it is relevant, integrate it where it contributes to the argument rather than appending a cluster of citations to satisfy a count. If it is not relevant, explain the mismatch with the paper’s question or claim. Avoid speculating about the reviewer’s identity or motives.

If a request appears coercive or inappropriate, raise the concern with the editor using the exact request and a factual explanation. Keep that communication professional and confidential. PLOS’s ethical publishing policy provides one example of explicit expectations for respectful editorial correspondence and routes for concerns. 4

A response should not make an unsupported accusation. It can say that the proposed references do not address the stated topic and ask the editor whether their inclusion is required. That keeps the issue tied to the manuscript and the editorial decision.

Prepare files that tell the same story

Follow the journal’s revision instructions rather than relying on what another journal required. PLOS One, for example, asks for a complete response, a clean manuscript and a marked copy; previously submitted supporting files can carry forward unless replaced. 2

Use clear version names and retain the submitted original. Check that the marked copy compares the correct versions. Remove private co-author comments from files intended for reviewers, while preserving the changes the journal needs to see.

Reconcile the response with the manuscript after pagination or line numbering is final. A location that was accurate before a late edit may no longer lead to the right passage. Use section and paragraph descriptions as well as line numbers when helpful.

Inspect supplements independently. Confirm that figure labels, file names, captions and links match the revised text. Open the submission system’s generated preview before completing the revision, because conversion can affect equations, tables or figure order.

Keep the editor’s copy readable

Use consistent headings for comments and responses, and preserve the numbering throughout the document. If a comment was split into several tasks internally, ensure the final response still answers the complete original request. Avoid making the editor reconstruct the relationship between two numbering systems.

Use emphasis sparingly and ensure changes are understandable without colour alone. A reviewer printing in black and white should still be able to distinguish the quoted comment, the authors’ answer and the revised manuscript passage. Remove internal status labels and unresolved editing notes before upload, while retaining the internal matrix for the team’s record.

Run a final contradiction check

Give a co-author the response and clean manuscript without explaining the changes verbally. Ask them to follow the most consequential responses to their locations and identify any claim that the manuscript does not support.

Check every “we have” statement against a completed action. Check every disagreement for a scientific reason. Check every new result against the underlying analysis. Check whether narrowing a conclusion in one section has left a broader claim in the abstract.

If the team cannot complete necessary work by the deadline, request an extension through the journal’s stated route before the deadline passes. Explain the remaining task and propose a realistic date. Do not upload an incomplete revision merely to create the appearance of compliance.

The editor may still disagree with your reasoning. A careful response does not guarantee acceptance. It gives the editor a clear basis for judging the revised work and leaves an accurate record of how the authors addressed the concerns.