Why papers get rejected, and what to do next

Papers get rejected for a small number of reasons, and telling them apart is the whole of deciding what to do next. A paper returned for fit needs a different journal by Friday. A paper rejected for a methodological gap needs work before it goes anywhere. A paper rejected on significance may need neither — only a venue whose threshold matches.

The common mistake is treating every rejection as the same event and resubmitting to the next journal down without reading what happened.

The weaknesses reviewers name have been counted. Georges Bordage content-analysed reviewers’ comments on 151 research manuscripts submitted to the 1997 and 1998 Research in Medical Education conference proceedings and published the ranked list in Academic Medicine (2001;76(9):889–896). Four peers reviewed each manuscript on average; among the 123 manuscripts that drew an “exclude” recommendation from at least one reviewer, each such reviewer wrote a mean of 8.1 reasons. Inappropriate or incomplete statistics ranked first. Overinterpretation of results ranked second. Neither is a statement that the science is wrong, and both are fixable in a manuscript that already exists.

The pages

• Desk rejected without review — the earliest rejection, and usually the most fixable.
• Rejected after review — what to do next — how to triage a review set into revise, resubmit, or restructure.
• How to write a response to reviewers — the document that decides more revisions than the experiments do.
• When to appeal a rejection — rarely worth it, occasionally correct, and there is a way to tell.

The causes, roughly in order of frequency

Rejections come from five causes: fit, significance below threshold, methodological gap, claims exceeding evidence, and presentation. Only the methodological gap normally requires new experiments. The other four are settled by where you sent the paper and how you wrote it, which is why the correct response to most rejections is a week of work rather than a year of it.

Fit

Fit. The paper is sound and the journal is wrong. Fastest to detect and fastest to fix.

A fit rejection is an editor’s judgement about readership, not about quality, and the International Committee of Medical Journal Editors states the basis for it plainly: “Editorial decisions should be based on the relevance of a manuscript to the journal and on the manuscript’s originality, quality, and contribution to evidence about important questions” (ICMJE Recommendations, section II.C). Relevance is listed first, and it is the only one of the four that a different journal can change without you touching the manuscript.

Fit failures are usually failures of shape rather than topic. A mechanistic journal returns a well-executed descriptive cohort study; a clinical journal returns a mouse study with a clinical framing; a specialist journal returns a paper whose comparator is from an adjacent field. The topic matched the aims and scope page and the paper still did not look like what the journal prints.

Detect fit by reading the last two issues rather than the aims and scope page, and match on shape — descriptive versus mechanistic, basic versus clinical, single-cohort versus multi-cohort — as well as subject. Twenty minutes of recent contents catches most of it; journal selection covers the procedure. The one trap is the polite significance rejection wearing fit’s clothing: “better suited to a more specialised journal” sometimes means the advance was too small for this venue, and moving sideways rather than down will collect the same letter again.

Significance below threshold

Significance below threshold. Sound work, insufficient advance for this venue. See novelty and scope objections.

A significance rejection says the work is correct and the increment is too small for this journal’s threshold, which is a competition statement rather than a quality statement. High-selectivity journals apply that threshold before review, so a significance rejection frequently arrives in days with no reviewer attached to it.

The trap is that significance is argued, not measured. Two papers with identical data can land on opposite sides of the same threshold because one names the question the field is stuck on and the other names the experiment that was run. Editors and reviewers read the abstract and the first paragraph of the discussion for the size of the claim; if the largest claim in the paper is the size of a single figure panel, the paper reads as incremental whatever else is in it.

Three responses work, in this order. Reframe around the question rather than the assay, so the advance is measured against what was previously unknown rather than against what was previously done. Move to a venue whose threshold matches the advance, which is the honest response most of the time and requires nothing but a resubmission. Add the one experiment that converts an observation into a claim, when that experiment exists and is quick — usually an orthogonal measurement rather than a new model system.

Never appeal a significance judgement. It asks an editor to weigh the same evidence differently, and appeals on that ground rarely succeed.

Methodological gap

Methodological gap. A design or analysis problem that undermines a central claim. The only category where new work is usually required — see methods and design objections.

A methodological gap is the only rejection category that is a statement that something is wrong, and it is also the category where authors most reliably misjudge the size of the remedy. The test is whether the central claim survives if the reviewer is correct. If the claim survives, the objection is about rigour of presentation and the fix is text. If the claim does not survive, either the experiment happens or the claim changes.

The gaps that recur are statistical before they are experimental. Bordage’s list puts inappropriate or incomplete statistics first and “sample too small or biased” fourth, and two concrete instances recur: a unit-of-analysis error — counting cells, sections or wells as independent replicates when the independent unit is the animal or the culture, covered at n not independent — and an uncorrected comparison count. Both are analysis defects in data you already have, which means both are fixable in the file rather than at the bench. Statistics objections sets out the ones that recur.

Genuine design gaps are narrower than the category suggests: a missing control that discriminates between two live explanations, a comparison confounded by batch or by time, an outcome measured only in the arm where it was expected. Those need new work. A reviewer asking for a knockout to accompany a knockdown, or a second cell line, is asking for orthogonal validation rather than a new study, and the smallest sufficient version of that request is usually acceptable if you say why it is sufficient.

Claims exceeding evidence

Claims exceeding evidence. The experiments support something smaller than the paper says. Frequently fixable by rewriting, and frequently misread by authors as a demand for more data.

Claims exceeding evidence is the rejection authors most often answer with the wrong instrument. The reviewer wrote that the data do not support the conclusion; the author books six months of experiments; the reviewer wanted the conclusion moved. Bordage ranked overinterpretation second among reasons for rejection, above every experimental complaint on the list.

The defect has a fixed shape. A correlational design is described in the discussion with causal verbs — drives, promotes, induces, protects against. A single model system is generalised to a disease. An effect measured in one sex, one strain, one centre or one time window is stated without its boundary. A title asserts what the abstract qualifies and the results only suggest. The gap is almost always between the results section, which is accurate, and the title, abstract and discussion, which are not.

Repair it by writing down the largest claim each figure supports on its own, then reading the title and the final paragraph of the discussion against that list. Every sentence above the line either acquires its qualifier or loses its verb. Boundaries belong in the results and the discussion as specifications, not in a defensive paragraph at the end; how to write a limitations section covers the difference between a stated boundary and an apology. The overclaim check exists for exactly this category, because it is the one where an author is least able to see their own manuscript.

Presentation

Presentation. Structure, clarity, figures, or a results section a reader cannot follow. Rarely the sole cause, often the aggravating one — a reviewer who cannot follow the argument assumes the argument is weak.

Presentation defects convert a borderline paper into a rejected one by changing what the reviewer believes about the science. Bordage’s ranked list places “text difficult to follow” fifth and “defective tables or figures” tenth, among reasons for rejection rather than among matters of taste — and those two sit alongside “insufficient data presented”, which is frequently the same defect seen from the reviewer’s side.

The instances that do the damage are specific. A results section that narrates the order experiments were performed rather than the order the argument needs. A figure whose panel labels do not match its legend. A number that appears in the abstract, the results and a table with three different values. Methods that omit the number of independent experiments. An abstract that states what was done without stating what was found — the single most expensive omission before review, because the editor’s decision is made on the title, the abstract and the figures; how to write an abstract sets out what belongs in each sentence.

Presentation is also the category most improved by one hostile read. Give the manuscript to someone in an adjacent field and ask them to state the main finding after ten minutes. If they cannot, a reviewer with four other manuscripts will not either.

How the cause maps to the next decision

The cause determines the next move, and the four moves are different in what they require and in how long they take.

The formal-requirement row belongs to desk rejection rather than to review, and it is the only row where resubmitting to the same journal is normally welcome.

What reviewers and editors actually write

Reviewers rarely name the category. They write the specific version, and the specific version is what tells you which category you are in.

“The manuscript addresses a question of limited interest to this journal’s readership.” That is fit. “The findings are convincing but represent an incremental advance on the authors’ previous work.” That is significance. “The n reported appears to be the number of cells rather than the number of independent biological replicates.” That is a methodological gap, and it is fixable in the analysis. “Sixteen comparisons are reported and no correction is applied; the three significant results are consistent with chance.” Also methodological, also fixable in the analysis. “The data demonstrate an association; the discussion describes a mechanism that was not tested.” That is claims exceeding evidence. “I was unable to determine from the Methods how many animals were used in each experiment.” That is presentation, and it is the sentence that precedes a reviewer deciding the work is careless.

Two of these carry more weight than the rest. A comment that appears in two independent reviews belongs in the must-address pile regardless of what you think of it: two people finding the same problem means the manuscript caused it, even if both are technically wrong, because something in the text created the misreading. And a comment phrased as a question about a number — “should Table 2 total 148 or 151?” — is a reviewer testing whether the paper can be trusted, not asking for arithmetic.

Reading the decision letter for the cause

The decision letter names the cause more often than authors credit, and the reliable signal is which part of the letter is specific.

An editor’s paragraph that is generic and a reviewer set that is specific means the reviewers decided. An editor’s paragraph that is specific about scope or priority, with reviews that are broadly positive, means the editor decided on significance and the reviews are not the reason. Reviews that disagree with each other, with an editorial paragraph that picks one, means the editor already told you which objection travels. Form letters with no reviews attached are decoded separately in desk rejected without review.

The letter also sets the boundary of what you may do next. Editors hold the decision: the ICMJE adopts the World Association of Medical Editors’ definition of editorial freedom, “which holds that editors-in-chief have full authority over the entire editorial content of their journal and the timing of publication of that content” (ICMJE Recommendations, section II.D). The same Recommendations state that “An editor can reject any article at any time before publication, including after acceptance if concerns arise about the integrity of the work” (ICMJE Recommendations, section II.C) — which is why an integrity query is answered immediately, completely, and with your institution informed, rather than argued.

Give it a week before deciding anything. Rejections are read badly on the day they arrive, and the pile-sorting in rejected after review works only on a second reading.

The one thing worth doing before resubmitting anywhere

Read the reviews for the objection that will recur. Reviewers at different journals are drawn from the same community and notice the same things. A criticism you decline to address does not disappear on resubmission; it arrives again, from someone with less patience.

That is the argument for fixing before moving rather than moving first: the second review is rarely gentler than the first.

The mechanism is small and specific. Subject-matter reviewer pools are small, the same names appear on the editorial boards of the journals immediately below the one that rejected you, and a reviewer who has already read the manuscript for another journal recognises it. Where the same objection is raised twice by different people at different venues, the objection is a property of the manuscript. Where an objection was raised once and never again, it was a property of that reviewer, and note it without acting on it.

Write the recurring objections down as a list before choosing the next journal, and choose the journal against the paper as it will be after the list is addressed — not as it was submitted.

What to do when the obvious remedy is unavailable

The remedy a reviewer names is frequently unavailable, and there is a defensible move in each case that is not silence.

The requested experiment cannot be done. The model does not exist, the reagent is discontinued, the cohort is closed, the patient samples are exhausted. State the reason in one sentence, offer the nearest available substitute, and say what the substitute does and does not establish. An orthogonal measurement in the material you have is worth more than a promise. If the experiment is impossible for a reason the reviewer did not know and the decision turned on it, that is one of the few grounds where an appeal works.

The objection is correct and the funding is gone. Narrow the claim to what the existing data support and publish the smaller finding at a matched venue. A correct, smaller paper is a publication; a contested, larger one is a queue of rejections.

Two reviewers contradict each other. If you are resubmitting elsewhere, neither is binding. Treat the contradiction as evidence that the manuscript was ambiguous enough to support both readings, and fix the ambiguity rather than picking a side. How to write a response to reviewers covers the version where you are revising rather than moving.

The methodological objection is fatal. The central claim does not survive and cannot be rescued. This is rare and it is real. The result is frequently still publishable as a negative or preliminary finding, and learning it from a reviewer is better than learning it after publication.

The rejecting journal offers a transfer to a sister title. Take it seriously. The reviews travel with the manuscript and the next round is usually much faster, provided the reason for rejection was fit or significance rather than a methodological gap that the sister journal’s reviewers will also find.

Detecting the cause before anyone rejects the paper

Four of the five causes are visible in a finished manuscript, and each has a check that takes under an hour.

Fit. Read the last two issues of the target journal and write down the shape of what they print. If your paper is not that shape, the aims and scope page will not save it.

Significance. Write the advance as one sentence beginning “Before this paper it was not known that…”. If the sentence is about a method having been applied rather than about something now known, the significance argument is not yet made. The study section lens tests that sentence the way a panel would.

Methodological gap. State the independent unit of analysis for every experiment, count the comparisons that produced every reported p-value, and check that the number in the abstract, the number in the table and the number in the model are the same number. Each of those three is checkable in an afternoon, and each is a defect in the file rather than at the bench.

Claims exceeding evidence. List the largest claim each figure supports alone, then read the title, the abstract’s final sentence and the discussion’s final paragraph against the list.

Presentation. Have someone outside your subfield read it and state the finding after ten minutes.

What journals publish about rejection rates, and what they do not

Journals do not uniformly publish desk-rejection or overall rejection rates, and the percentages that circulate in advice articles are usually offered without a journal name, a year, or a denominator. Treat any rejection rate presented that way as unsourced, including the ones that sound authoritative.

What is documented is the ranked content of reviewer criticism rather than its frequency across the literature. Bordage’s 151-manuscript analysis in Academic Medicine (2001;76(9):889–896) is a published enumeration of what reviewers wrote, and its top two entries — inappropriate or incomplete statistics, and overinterpretation of results — are both defects of a written manuscript rather than of an experiment. Individual journals publish their own acceptance figures inconsistently and define them differently, so two published rates are rarely comparable: one journal counts every submission in the denominator, another counts only those sent for review.

The practical consequence is that the base rate is not the useful number. The rejection rate of the journal that rejected you changes nothing you would do next; whether your rejection was fit, significance, method, claim or presentation changes everything.

Before submission

Most of these causes are visible in a manuscript before anyone rejects it. PerfectPaper’s standing review covers methodology, statistics, evidence, structure and journal fit; custom reviewer agents add the field-specific conventions a general reviewer misses. The two most useful pre-submission passes for this list are the overclaim check, for claims exceeding evidence, and the study section lens, for significance and framing.

PerfectPaper reads a manuscript against the five causes of rejection before an editor does, and names which of them a submitted draft would trigger. Peer review remains the editor’s and the reviewers’ decision; what changes is how much of the avoidable half is still in the file when they see it.

Related

Desk rejected without review · Rejected after review · When to appeal · Novelty objections · Methods objections · Journal selection

Review my manuscript

Frequently asked questions

Why do papers get rejected from journals?

Papers get rejected for five causes: fit, significance below the venue’s threshold, a methodological gap, claims exceeding the evidence, and presentation. Only the methodological gap normally requires new experiments; the other four are decided by venue choice and by writing.

What reasons do reviewers actually give for rejecting a manuscript?

Bordage’s content analysis of reviewers’ comments on 151 manuscripts, published in Academic Medicine (2001;76(9):889–896), ranked inappropriate or incomplete statistics first, overinterpretation of results second, inappropriate or suboptimal instrumentation third, a sample too small or biased fourth, and text difficult to follow fifth.

What is the most common reason a manuscript is rejected?

No single figure covers every journal, and rejection rates are rarely published with a denominator. Where reviewer comments have been counted, statistics and overinterpretation lead the list — and both are defects of a manuscript that already exists rather than of the experiments in it.

Why do papers get rejected without being sent for review?

Desk rejection is an editor’s judgement made on the title, abstract, figures and formal requirements alone: relevance to the readership, an advance below the journal’s threshold, or a missing statement or identifier. Desk rejected without review decodes the form letter.

Why was my paper rejected when the reviewers said the work was sound?

A sound-but-rejected paper is a fit or significance decision rather than a quality one. Both say the work is correct and belongs somewhere else — fit because the readership is wrong, significance because the advance is smaller than this venue prints.

Why does the same paper get rejected by several journals in a row?

The objection is usually a property of the manuscript rather than of one reviewer. Subject-matter reviewer pools overlap between neighbouring journals, so a criticism you decline to address tends to arrive again from someone with less patience.