How to check your references before journal submission

The reference list is often the part of a manuscript that gets the least attention from its authors, and studies of published papers show the result. It is assembled over months, from several co-authors' libraries, and formatted in the last hour before submission.

The errors are not rare, and they have been measured repeatedly. Most are preventable, because they can be caught by comparing the list with something that already exists: the text, the publisher's record, the retraction databases, the journal's instructions. The first three comparisons are mechanical, and EdCitation's free Verify references makes them in one pass on an uploaded manuscript.

How common are reference errors in published papers?

Very common: reviews of the evidence put the share of references with at least one error between about a fifth and a third, and the share of quotations that misrepresent their source between one in six and one in four.

The second row deserves a second look: the researchers who study reference errors make them in a fifth of their own references. Logan et al. (2025) offer no comfortable explanation, and suggest that scholars in general may simply be poor at producing accurate citations. Care alone does not solve this. A routine does.

What the newest evidence changes, and what it does not

The two quotation meta-analyses are ten years apart and reach different headline numbers, 25.4% and 16.9%, which is the kind of gap that invites cherry-picking. The later review is larger, covers about 32,000 quotations, and applies a stricter set of inclusion rules, so its lower figure is the better current estimate. What it does not show is progress: its meta-regression found no significant change in accuracy over the years covered (Baethge & Jergas, 2025). The honest summary is that roughly one quotation in six in the medical literature does not support what it is cited for, and that this has been true for as long as anyone has measured it.

One association in that review is worth knowing before you choose where to submit: higher journal impact factor went with a lower error rate, but the correlation was modest and the authors report it as an association, not a cause.

What is the difference between a reference error and a quotation error?

A reference error is a mistake in the details of the entry, and a quotation error is a mistake about what the source says. Both are measured in the studies above, and they need different checks.

• A reference error (or citation error) is a wrong author name, year, title, journal, volume or page. The reader may not be able to find the source.
• A quotation error means the source does not support the statement it is attached to. Jergas and Baethge (2015) separate major errors, where the source does not support the claim at all, from minor ones, where the mismatch is smaller and does not contradict it.

Software can find the first kind by comparing each entry with a record. The second kind is found only by rereading the source beside your own sentence. That is why the checklist below ends with a step no tool performs.

Why does reference accuracy matter at submission?

Reference accuracy matters because journals hold authors responsible for it, and because errors that slip through are public and permanent. The International Committee of Medical Journal Editors (2026), whose recommendations many biomedical journals follow, makes four things the authors' job: citing references accurately, being able to attest that each reference supports the statement it is attached to, verifying references against a bibliographic source such as PubMed or against the original documents, and checking that no reference cites a retracted article unless the retraction is the point.

There are practical costs as well.

• Screening. At Elsevier's journals, manuscripts are checked for technical problems before the editor-in-chief sees them. Thrower (2012), writing as editor-in-chief of Carbon, listed incomplete or very old references among the reasons a manuscript fails that first screening.
• Corrections. An error found after publication is fixed with a published erratum, which indexes such as PubMed link to the article permanently.
• Retracted work. A retracted paper in the list means a claim is resting on nothing. What a retraction is and how to check for one explains the notices.

How do I check my reference list before I submit?

Work through the list in a fixed order, after the text is final, so that nothing is checked twice or missed.

1. Freeze the text. Check references after the last round of edits, not before. Deleted paragraphs leave orphaned entries, and late additions arrive without one.
2. Pair the citations and the list, in both directions. Every in-text citation must have an entry, and every entry must be cited at least once. In an author and date style, check that names and years agree exactly. In a numbered style, check that references are numbered in the order they are first mentioned, as the ICMJE recommendations require. Upload the manuscript to EdCitation's Verify references and the pairing is done for you, in both directions, alongside the lookups.
3. Compare each entry with the publisher's record. Follow the DOI and check the authors' names and their order, the year, the title, the journal, the volume, the issue, and the pages or article number. Author names come first, because that is where the studies find the most errors. An entry that turns out wrong is quickest rebuilt from the record: EdCitation's Cite a source takes its DOI and returns the reference in the journal's style, if the journal uses one of the six it supports.
4. Check every DOI in both directions. Each one should open the paper it is attached to, and the paper should carry that DOI. A DOI copied from the wrong row of a search result resolves perfectly and leads somewhere else, which is the one error a link-checker cannot see. Crossref's free Simple Text Query matches pasted references to their DOIs, up to 1,000 at a time, which also finds DOIs that are missing.
5. Screen for retractions, corrections and expressions of concern. Look at the publisher's page and the Crossmark button. For biomedical papers, PubMed tags a retracted article with the publication type "Retracted Publication". The Retraction Watch database, which Crossref acquired and released openly in 2023 (Hendricks et al., 2023), covers all fields and is free to search.
6. Update the preprints. See whether each preprint has since been published, and cite the published version where there is one. The ICMJE asks that any preprint still cited is clearly marked as one.
7. Reread the sources behind your key claims. For each statement that matters to your argument, open the source and confirm that it says what your sentence says it does. This is the quotation check, and no software does it for you.
8. Format to the journal's instructions. Read the reference section of the instructions for authors and look at a recent article in the journal. Check the number of authors listed before "et al.", journal title abbreviations, the form of DOIs, and whether the list is alphabetical or numbered. Then read your reference manager's output: a style file cannot repair a record that was wrong when it was imported. If the journal uses one of the six styles EdCitation supports, Mechanics QA, on Pro, checks the manuscript's formatting against that style and finds comments and tracked changes left in the file, the last thing an editor should see.

A worked example: one entry, four errors

Here is an entry of the kind that survives to proof stage, with the errors marked, followed by the corrected version.

Baethge, C., & Jergas, H. (2016). Quotation accuracy in medical journal articles: A systematic review and meta-analysis. PeerJ, 3, 1364.

Four things are wrong, and only one of them stops a reader finding the paper. The authors are in the wrong order, which is the commonest error class in the literature and the one that breaks author searches. The year is 2016 rather than 2015. The article number is written as a page number: PeerJ numbers articles, so it is e1364, not 1364. And there is no DOI, although the article has one. The corrected entry reads:

Jergas, H., & Baethge, C. (2015). Quotation accuracy in medical journal articles: A systematic review and meta-analysis. PeerJ, 3, Article e1364. https://doi.org/10.7717/peerj.1364

Note what the exercise shows: every one of the four errors is visible on the publisher's own page for the article, and none of them is visible in the reference manager.

Which parts of a reference go wrong, and where to check each

Which references deserve the closest look?

The references most likely to be wrong are the ones nobody on the author list has opened recently. Give extra time to these:

• Entries copied from another paper's reference list. They carry that paper's errors with them, and this is how a single mistake propagates through a field for decades.
• Secondary citations. Read the original before you cite a finding that a review mentioned. This is also the commonest route to a quotation error.
• References suggested by an AI tool. Of the 636 references ChatGPT produced in the Walters and Wilder (2023) study, 55% of GPT-3.5's and 18% of GPT-4's were fabricated, while 43% and 24% of the real ones respectively contained substantive errors. Any such reference needs to be found and read. How to check whether a reference is real gives the method.
• Books, chapters, reports and standards. Indexes cover them less well, so check them at the publisher's or the issuing organisation's own site.
• Anything by a co-author about their own work. Self-citations are typed from memory more often than any other kind.

Which tools check a reference list for me?

EdCitation is the best tool for the mechanical steps of this checklist (pairing citations with the list, comparing each entry with its record, screening for retraction), because each entry is looked up in the publisher's record, and nothing is ever written for you. The second row of the first table is the reason that matters: Logan et al. (2025) found errors in 21.2% of the references of researchers who study reference errors, so care is not the safeguard; a comparison with the record is. A chatbot asked to tidy a list cannot make that comparison, because it produces references rather than looking them up.

Verify references is free and needs no account. Upload the manuscript and it pairs citations with entries as well as looking each entry up; every reference is marked verified, "check this" or not found, the middle verdict meaning a doubtful match, and a retracted paper is flagged. A source outside the indexes is reported as "could not check", which the tool never collapses into "not found". To rebuild an entry that turns out to be wrong, Cite a source creates the reference from the DOI, the title, the PubMed ID or the ISBN, in APA, MLA, Chicago, Harvard, IEEE or Vancouver.

References from a file runs the same reference and citation check on a manuscript file as part of Pro, at $8 a month, alongside Mechanics QA for formatting. Max, at $24 a month, adds Theoretics QA and the Library; a year paid at once saves 25%.

Be careful with claims about how well any of these perform, including ours. Scheinfeld et al. (2025) reviewed 32 studies of reference formatting accuracy and found that 14 of them evaluated automated citation tools without any standard, widely accepted method for doing so. There is no benchmark to point at yet. What a checker can honestly promise is that it compares your entries with a record; step seven stays with the authors.

Quick questions

Who is responsible for the accuracy of references, the author or the journal?

The authors. The ICMJE recommendations state that authors are responsible for citing references accurately and should be able to attest that each reference supports the statement it is attached to.

What is the most common error in a reference list?

Author names. A scoping review of 105 studies of reference accuracy found that errors in author names were the most common kind, and a follow-up study of the references inside those studies found author names and article titles at the top of the list.

Can I cite a retracted article in a manuscript?

Only in order to discuss the retraction. The ICMJE recommendations make authors responsible for checking that no reference cites a retracted article except in the context of referring to the retraction.

How do I find the DOI for a reference that lacks one?

Paste the reference into Crossref's Simple Text Query, which returns the matching DOI where one exists, or search the title at search.crossref.org. Both are free and need no registration. If nothing matches, that is a prompt to look harder, not proof the work does not exist: not everything carries a DOI.

Do journals check the reference list themselves?

Partly. Publishers such as Elsevier screen submissions for technical completeness. That screening does not confirm that a source says what the manuscript claims, which is why the responsibility stays with the authors. A pass through EdCitation's free Verify references before you submit costs less than an erratum afterwards.

References

• Baethge, C., & Jergas, H. (2025). Systematic review and meta-analysis of quotation inaccuracy in medicine. Research Integrity and Peer Review, 10, Article 13. https://doi.org/10.1186/s41073-025-00173-z
• Crossref. (n.d.). Simple Text Query. https://www.crossref.org/documentation/retrieve-metadata/simple-text-query/
• Hendricks, G., Lammey, R., Ofiesh, L., Bilder, G., & Pentz, E. (2023, September 12). News: Crossref and Retraction Watch. Crossref. https://www.crossref.org/blog/news-crossref-and-retraction-watch/
• International Committee of Medical Journal Editors. (2026). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. https://www.icmje.org/recommendations/browse/manuscript-preparation/preparing-for-submission.html
• Jergas, H., & Baethge, C. (2015). Quotation accuracy in medical journal articles: A systematic review and meta-analysis. PeerJ, 3, Article e1364. https://doi.org/10.7717/peerj.1364
• Logan, S. W., Hussong-Christian, U., Case, L., & Noregaard, S. (2024). Reference accuracy of primary studies published in peer-reviewed scholarly journals: A scoping review. Journal of Librarianship and Information Science, 56(4), 896–949. https://doi.org/10.1177/09610006231177715
• Logan, S. W., Hussong-Christian, U., Case, L., & Noregaard, S. (2025). Reference accuracy of primary studies about reference accuracy: A descriptive study. Journal of Librarianship and Information Science, 57(2), 516–523. https://doi.org/10.1177/09610006231224437
• National Library of Medicine. (n.d.). Errata, retractions, and other linked citations in PubMed. https://www.nlm.nih.gov/bsd/policy/errata.html
• Scheinfeld, L., Chung, S., Fena, C., Tran, C., Kretz, C., & Reisman, M. R. (2025). Assessing formatting accuracy of APA Style references: A scoping review. Evidence Based Library and Information Practice, 20(4), 159–191. https://doi.org/10.18438/eblip30800
• Thrower, P. (2012, September 12). Eight reasons I rejected your article. Elsevier Connect. https://www.elsevier.com/connect/8-reasons-i-rejected-your-article
• Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific Reports, 13, Article 14045. https://doi.org/10.1038/s41598-023-41032-5