Manuscript Review Template: A Referee Report Structure With a Worked Example
Every referee report template has the same five boxes. The hard part is deciding which box a problem goes in. Here is the template, a one-question test for major versus minor, and a filled-in report you can copy the shape of.
While you wait
Waiting on a decision? Get your next move ready.
The wait is out of your hands; the next move isn't. Scan your next manuscript free, or run this paper through the scan to see what reviewers typically push back on, so the revision response is ready when the decision lands.
How to use this page well
These pages work best when they behave like tools, not essays. Use the quick structure first, then apply it to the exact journal and manuscript situation.
Question | What to do |
|---|---|
Use this page for | A working artifact you can actually apply to the manuscript or response package. |
Start with | Fill the template with real manuscript-specific details instead of leaving it generic. |
Common mistake | Copying the structure without tailoring the logic to the actual submission. |
Best next step | Use the artifact once, then cut anything that does not affect the decision. |
Quick answer: A manuscript review template has five parts: a short summary of what the paper claims, an overall assessment of whether the evidence carries that claim, numbered major comments, numbered minor comments, and a recommendation. Most journals add a sixth box, confidential comments to the editor. The boxes are standard everywhere. What separates a useful report from a weak one is triage, and one question does most of it: could fixing this change a conclusion, a figure, or the decision? If yes, it's major. If not, it's minor.
Run the free readiness scan on your own manuscript to see the top reviewer risk before you ask a colleague for a full read, or copy the template below.
Evidence basis: We looked at what ranks for this search on September 22, 2026 (a journal reviewer template, publisher guides, funder-platform blogs and generic form builders) and compared their structure with the report our own pre-submission review produces. This page quotes no journal's rules. The template and the triage test below are our editorial synthesis. The worked example is invented for this page. The journal your review is for always wins: if its reviewer form has fixed questions, answer those, and use this structure inside the free-text boxes.
The template
Copy this into a blank document before you open the manuscript a second time.
MANUSCRIPT: [ID / short title] RECOMMENDATION: [editor field only]
1. SUMMARY (2 to 4 sentences, your words, not the abstract's)
The authors claim ___.
They test it by ___ in ___ (system, sample, n).
The main evidence is Figure/Table ___.
2. OVERALL ASSESSMENT (1 short paragraph)
Does the evidence support the central claim as stated?
Yes / Partly (which part holds, which doesn't) / No.
The one change that would most improve the paper: ___.
3. MAJOR COMMENTS (numbered, most serious first)
M1. Where: [section, figure, line]
Problem: ___
Why it matters for the conclusion: ___
What would resolve it: [analysis, control, reframing, or softer claim]
4. MINOR COMMENTS (numbered, in manuscript order)
m1. [line / figure]: ___
5. CONFIDENTIAL COMMENTS TO THE EDITOR (authors never see this)
Conflicts, suspected integrity issues, your confidence in your own
expertise on each part, and the recommendation with its reasons.Two details in that template do more work than they look like they do.
The "Where" line on every major comment. Authors answer comments they can find. "The statistics are weak" produces a defensive paragraph in the response letter. "Figure 3B compares six groups with repeated t-tests and no correction" produces a reanalysis.
The "What would resolve it" line. It forces you to check that your request is proportionate. If the only thing that resolves the comment is a two-year follow-up study, you're really saying the claim is too strong, and the fair request is a narrower claim, not the study.
Major or minor: the one-question test
Most reports go wrong in the sorting, not the writing. In practice, reviewers tend to rank a problem by how much it annoyed them. Editors read the major list as the price of acceptance. Those are different scales.
Sort by consequence instead:
Fixing it could | It is | Example |
|---|---|---|
Change the central conclusion | Major | A missing control that could explain the main effect on its own |
Change a figure or a reported number | Major | Technical replicates counted as independent n, so the reported P value is too small |
Change the editor's decision, even with the science intact | Major | The claim is framed as causal throughout, and the design only supports association |
Change how easily a reader follows it, but not what it says | Minor | Inconsistent abbreviations, a missing scale bar value, a figure panel out of order |
Change nothing a reader relies on | Leave it out, or one line at the end | Your preferred citation style, a synonym you like better |
The middle row is the one reviewers underuse. Plenty of solid experiments get sunk by one sentence in the abstract that promises more than the data give. That's major even though it takes the authors ten minutes to fix, because an editor reading the abstract decides on the claim as written.
A useful count: if you have more than five major comments, reread them. Usually two are the same problem seen from different sections, and one is a minor point you felt strongly about. Three sharp major comments get acted on. Nine blur together.
A filled-in example
Here's the template filled in for an invented manuscript. The paper, the numbers and the flaws were made up to show the shape. None of it describes a real submission.
Summary. The authors claim that compound K reduces liver fibrosis in mice by blocking TGF-β signalling in hepatic stellate cells. They test it in a carbon tetrachloride model (n = 8 per group), measure collagen by Sirius Red staining and hydroxyproline, and show reduced SMAD2 phosphorylation in isolated stellate cells. The main evidence is Figures 2 and 4.
Overall assessment. The anti-fibrotic effect in Figure 2 is convincing. The mechanism is not yet shown: the paper demonstrates that compound K lowers pSMAD2, not that this is how it reduces fibrosis. Either a rescue experiment or a narrower claim would fix the gap.
M1. Where: Figure 4, Discussion paragraph 2, title. Problem: reduced pSMAD2 is correlated with reduced fibrosis, but no experiment shows the pathway is required for the effect. Why it matters: the title and abstract state a mechanism. Resolve by: a constitutively active TGF-β receptor rescue in stellate cells, or retitling to report the association.
M2. Where: Figure 2C, Methods (Statistics). Problem: the legend gives n = 8, but the Methods say three fields per liver were quantified, and the degrees of freedom in the reported test suggest fields were analysed as independent. Why it matters: if n is 24 fields, the P value is overstated. Resolve by: reanalysing with one value per mouse, or a mixed model with mouse as a random effect.
m1. Figure 3 has no scale bar value in the legend. m2. "HSC" is used before it is defined (Introduction, line 42).
Notice what M2 does. It doesn't accuse anyone. It lays out the arithmetic that made you suspicious and offers two fixes, one of which costs the authors an afternoon. Comments like that get fixed without a fight.
What goes to the editor, and what goes to the authors
Put it in comments to the authors | Put it only in confidential comments to the editor |
|---|---|
Every scientific problem, with its location and a proposed fix | Your recommendation, with a sentence on why |
Requests for data, code, or uncropped images | Suspected duplicated or manipulated images, plagiarism, or ethics problems |
Questions you genuinely can't answer from the text | A conflict of interest you noticed after accepting |
Tone you'd accept if the report were signed | Which parts of the paper fall outside your expertise |
The last row is the most useful thing a reviewer can tell an editor and the one most often skipped. "I can assess the imaging and the mouse work; I'm not qualified to judge the proteomics" tells the editor exactly where a second opinion is needed.
One boundary applies before you start. A manuscript under review is confidential, and it isn't yours to share. Read the journal's reviewer guidelines before you paste any part of it into a chatbot or any other AI tool, including ours. Manusights is built for authors checking their own manuscripts.
Reviewing your own manuscript with the same template
The template is at least as useful on the author's side of the desk. Most researchers read their own draft as the person who ran the experiments, which means they read what they meant rather than what's on the page. A referee structure breaks that.
Try it in this order:
- Write the summary box from memory, without opening the abstract. Then compare. If your summary and the abstract claim different things, the paper doesn't yet say what you think it says, and that's the first fix.
- List every weakness you already know about. Every lab has them: the control you didn't have time for, the cohort that's smaller than you'd like. Run each through the one-question test. The ones that come out major belong in the Limitations paragraph or in a narrower claim, before a referee raises them for you.
- Hand the blank template to a co-author who didn't write the Results. Ask for major comments only. Most internal reads come back as tracked-changes grammar edits because nobody asked for anything else.
- Check the "Where" for each of your own claims. Every sentence in the abstract that states a result should map to a specific figure or table. A claim with no "Where" is the one a referee will find.
Readiness check
While you wait, scan your next manuscript.
The scan takes about 1-2 minutes. Use the result to decide whether to revise before the decision comes back.
What our pre-submission review report contains
Our AI engine was trained by 35+ current top-tier-journal reviewers, and its report follows the same logic as the template above: find the claim, test the evidence, rank the problems by consequence. The details differ by tier, and it's worth knowing exactly what each one gives you.
The free readiness scan (about 1-2 minutes, no card) returns a submission-risk signal and a preview of the top blocker. Think of it as the overall-assessment box and M1, nothing more.
The $39 Full Review scores the manuscript on seven dimensions, from originality and methodological soundness to whether the claims are supported and how well prior work is placed. It lists what the data package does well and where it falls short, and each weakness has to name the figure or panel it refers to and carry a concrete suggestion. Revisions come as a short lettered list, each tagged Essential or Recommended, which is our version of the major-versus-minor split. References are checked against Crossref, PubMed and Retraction Watch. One deliberate choice: there is no minor-comments appendix. Typos and label fixes are what your co-authors are for.
The $99 Submission-Ready Dossier is the closest thing to the referee report on this page. Four simulated referees read the paper for rigor, novelty, domain fit and reproducibility. Every issue they raise is rated critical, major or minor, tied to a verbatim quote from your manuscript (checked against your text, so every "Where" points at words that are really on the page), and paired with a suggested action.
The Dossier also checks CONSORT, STROBE, PRISMA 2020 or ARRIVE reporting where they apply, and proposes a first-choice journal with safer backup venues.
None of this replaces an expert in your subfield who knows which result is already in someone else's preprint. What it gives you is a structured, referee-style report while the manuscript is still yours to change.
In our pre-submission review work, four major comments keep coming back
In our pre-submission review work with research manuscripts, the major-comments box starts to look familiar. Each of the four below is a named failure pattern we see often enough to put in the major box every time, because each one can change a conclusion or a decision. Check your own manuscript for them before anyone else fills in that box.
1. The claim that drifts. The abstract says "drives", the Results say "is associated with", and the Discussion says "may contribute to". Three strengths of the same claim in one manuscript. The referee picks the strongest version, tests it against the weakest evidence, and writes M1. Fix it by choosing the claim your figures support and using those words in all three places.
2. The n that isn't. Fields, wells, slices or technical repeats get counted as independent samples, so the statistical analysis reports more certainty than the design allows. It's the M2 in our worked example, and it's a major comment even when the effect is real, because every P value built on it moves.
3. The comparison nobody ran. The treated group is compared with baseline instead of with a time-matched control, or a new method is benchmarked against an outdated one. The fix is usually one extra panel or one extra row in a table. Without it, the headline result has an obvious alternative explanation, and finding those is the referee's whole job.
4. The orphan sentence. A sentence in the abstract or conclusion that no figure, table or reference supports. It's often the most quotable line in the manuscript, which is why it survives drafting. A reviewer filling in the "Where" line for that claim finds nothing to point at.
None of these four is exotic. What makes them expensive is timing: caught by a co-author with this template, each costs an afternoon; caught by a referee, each costs a revision round. That difference is the whole case for reviewing your own manuscript before you submit it.
When a report falls short
Think twice before sending your report if any of these is true:
- The summary paraphrases the abstract. Then you haven't shown the editor you understood the paper, and you may have missed where the claim quietly shifts between abstract and Discussion.
- Every comment asks for a new experiment. Some should ask for a softer claim instead. A request that would take a year is often a rejection disguised as a revision.
- The major list has no "Where" lines. Authors will answer the comments they can locate and argue with the rest.
- The tone would embarrass you if the report were signed. Some journals publish reports or reveal reviewer names. Write every report as if yours will be one of them.
- You reviewed the paper you'd have written. "The authors should have used approach Y" is fair only when approach X can't answer the question the paper asks.
These are the reviewer guides that appeared in the live results for this search when the page was written. This page quotes none of them; use your journal's own reviewer form when it has one.
Frequently asked questions
Five parts: a two-to-four sentence summary of what the paper claims and how it tests the claim, an overall assessment of whether the evidence supports that claim, numbered major comments, numbered minor comments, and a recommendation. Most journals also give reviewers a separate box for confidential comments to the editor, which the authors never see.
Ask whether fixing it could change a conclusion, a figure, or the editor's decision. If yes, it is major. If the paper's claims would stand unchanged after the fix, it is minor. A missing control that could explain the main result is major; a mislabelled axis is minor, even if it irritates you more.
Long enough that each major comment names the problem, points to where it sits in the manuscript, says why it matters for the conclusion, and says what would resolve it. For a typical research article that is often one to two pages. A report of forty minor comments and no clear priority is harder for authors to act on than a short one with three well-argued major points.
Check the journal's reviewer instructions. Many submission systems ask for the recommendation in a separate field or in the confidential comments to the editor, because the editor makes the decision and may weigh several reports that disagree. When in doubt, keep an explicit accept or reject verdict out of the text the authors will read.
Yes, and it works better than rereading. Write the summary of your own paper without looking at the abstract, then compare the two. If they differ, the paper does not yet say what you think it says. Then run the major-versus-minor test on every weakness you already know about.
Final step
Done interpreting the status? Put the wait to work.
The decision will arrive on the journal's clock. What you control is what's next: scan your next manuscript free, or run this paper through the scan so the likely reviewer pushback is mapped before the revision request lands.
Free scan, no card needed.
Private API processing. Your manuscript is not used to train models.
Re-read the status guidePut the guidance to work
Move from draft advice to a practical manuscript pass.
Choose the reporting framework first, then use the final checklist when the manuscript and submission files are ready to travel together.