
Best Tools for Verifying Scholarly References (Free, 2026)
By Samuel de Bruyn 14 min read Updated
The best free tools for verifying scholarly references are Crossref’s Simple Text Query, OpenAlex, Zotero’s retraction flags and Litrevu’s reference checker. Crossref matches up to 1,000 pasted references to DOIs in one submission, OpenAlex confirms a single title or DOI against an open catalogue, Zotero flags retracted papers already in your library, and Litrevu’s reference checker tests up to 30 references at a time for existence, wrong details, a DOI that belongs to another paper, and retraction.
None of those four tools will tell you whether a paper says what your sentence claims it says. That check is still manual, and the ICMJE Recommendations and university AI guidance put it on the author, not on the tool.
What is reference checking? Reference checking means confirming two separate things about every entry in a bibliography: that the work exists as described, and that the details recorded for it are correct. A fabricated citation describes a paper that was never published; an erroneous citation points at a real paper but gets the author, year, journal, volume or page numbers wrong.
One declaration before the list: I built Litrevu, one of the four tools, so weigh what I say about it accordingly. Every row below sticks to what each tool’s own pages state, including ours, as read in November 2026.
What are the best tools for verifying scholarly references?
Four free tools cover the reference checking job between them: Crossref, OpenAlex, Zotero and Litrevu’s reference checker. They overlap less than their descriptions suggest, because each one answers a different question about a reference, and the table shows which.
| Tool | What you give it | What it verifies | What it will not catch | Cost |
|---|---|---|---|---|
| Crossref (Simple Text Query, plus the retraction filter in its REST API) | Up to 1,000 references pasted as text, or a programmatic query | Whether a DOI exists for the reference as written, and which DOI; through the API filter, every retraction notice Crossref holds from the Retraction Watch database | References with no DOI, such as many books and theses, and whether the paper supports your sentence | Free |
| OpenAlex | A title, DOI or author query, on the site or through the API | Whether the work is in OpenAlex’s free and open catalogue of about half a billion scholarly works, with its metadata | A whole pasted list in one pass, and whether the paper supports your sentence | Free data under CC0; every account gets a free daily API budget |
| Zotero retraction flags | Your existing Zotero library | Whether an item you have saved has been retracted, matched by DOI or PMID against Retraction Watch data | Items with no DOI or PMID, and references that are not in Zotero | Free |
| Litrevu reference checker | Up to 30 references pasted as text, in any style | Existence; the year, first author, volume, first page and DOI against the record; whether the DOI belongs to that paper; retractions and expressions of concern | Whether the paper supports your sentence, works that none of its sources hold, and lists longer than 30 per check | Free, no sign-up |
Crossref’s Simple Text Query says it provides "space to match 1,000 references per submission", and tells you that "To check the accuracy of a reference, click on the link that is returned with the reference." OpenAlex describes itself as "a free and open catalog of the world’s research" that indexes "half a billion scholarly works", and its pricing page says the data is "released under a CC0 public-domain license". Litrevu’s reference checker looks in Crossref and OpenAlex, then in CORE’s university repositories and on the page a reference links to, and its page states that "No AI model is involved: fixed rules decide each verdict."
What is the fastest way to check whether a reference is real?
Paste the whole reference list into a tool that matches each entry against Crossref, the registry publishers deposit DOIs into, then open whatever it flags. Crossref’s Simple Text Query takes up to 1,000 references in one submission and returns the matched DOI for each one, and Litrevu’s reference checker takes up to 30 references per check with no account.
A reference that comes back with no match is not automatically fake. Book chapters, theses, conference papers and older articles often have no DOI at all, so a "no match" result on those entries is a signal to search by hand, not a verdict. A reference that comes back matched to a different paper from the one you typed is the result worth slowing down for, because a mismatch usually means the author, year or title in your bibliography is wrong, or that the reference was invented around a real DOI.
How do you find references that are never cited in the text?
To find references that are never cited in the text, match every in-text citation to an entry in the reference list, then match every entry in the list back to the text: any entry that no citation points to is uncited. Litrevu’s free In-Text Citation Checker runs both passes on a pasted essay or a .docx file, and lists "References you never cite", which its page defines as "An entry that nothing in your text points to."
The In-Text Citation Checker also flags the opposite slip, a citation with no entry in the list, plus years and names that disagree between text and list (citing Braun and Clarke 2008 when the list says 2006) and citations that fit more than one entry. The check runs in the browser, so the essay is not sent to Litrevu, and the tool’s page says its matching rules were checked against APA Style, Cite Them Right Harvard, the Chicago Manual of Style, the MLA Style Center, the IEEE Reference Guide and the ICMJE Recommendations. Use it to find references never cited in the text before you run the reference list itself through an existence check.
Each problem in a reference list has a different tool that catches it:
| Problem in your references | Which tool catches it |
|---|---|
| A reference that does not exist | Crossref Simple Text Query (no match) or Litrevu’s reference checker ("We could not find this"), then a manual catalogue search |
| Wrong year, first author, volume or page | Litrevu’s reference checker ("Found, but details differ") |
| A real DOI attached to the wrong paper | Litrevu’s reference checker ("The DOI belongs to a different paper") |
| A retracted paper | Zotero (saved items with a DOI or PMID), Crossref’s retraction filter, Litrevu’s reference checker |
| An entry never cited in the text | Litrevu’s In-Text Citation Checker |
| A citation with no entry, or a year that disagrees with the list | Litrevu’s In-Text Citation Checker |
| A paper that does not support your sentence | No checker; open the paper and read the passage |
Why do real-looking references turn out not to exist?
Real-looking references turn out not to exist because a language model producing a citation from memory is assembling a plausible string, not retrieving a record. Walters and Wilder measured fabricated citations directly in Scientific Reports (2023, volume 13, article 14045, doi 10.1038/s41598-023-41032-5): across 636 citations from 84 short literature reviews on 42 topics, "55% of the GPT-3.5 citations but just 18% of the GPT-4 citations are fabricated."
Two cautions apply to carrying the Walters and Wilder percentages into 2026. The study tested GPT-3.5 and GPT-4 in 2023, and the models students use now are different, so the figures are not current rates. The finding that has held up is about appearance: a fabricated citation is formatted exactly as well as a real one, so nothing in the look of a reference list tells you which entries will survive a check.
Is it enough to check that the DOI resolves?
No. A resolving DOI proves that something was registered under that DOI, and says nothing about whether your entry describes that work correctly. Walters and Wilder found in Scientific Reports (2023) that "43% of the real (non-fabricated) GPT-3.5 citations but just 24% of the real GPT-4 citations include substantive citation errors", so a citation can point at a genuine paper and still carry the wrong year, journal or author order.
A real DOI can also be attached to the wrong paper altogether, which is why Litrevu’s reference checker opens the record the DOI is registered to and compares its title, first author and year with the reference as written, rather than stopping at "the DOI works". Nor does a resolving DOI show that a work was peer reviewed, because preprints, theses and datasets get DOIs too. Litrevu’s peer review checker takes one DOI, link or title and shows what kind of work it is, such as a journal article or a preprint.
The standard journals apply goes further than existence. The ICMJE Recommendations state that "Authors are responsible for citing references accurately and should be able to attest that the references cited support the associated statement", and that "To minimize bibliographic citation errors, references should be verified using either a bibliographic source, such as PubMed, or original sources." Attesting that a reference supports a statement is not something a DOI lookup can do; it means opening the paper.
How do I check whether something I cited has been retracted?
Check the reference against the Retraction Watch database, which Crossref now carries and which is "updated every working day by Retraction Watch", according to Crossref’s documentation. Crossref exposes the data two ways: a REST API filter, documented as api.crossref.org/v1/works?filter=update-type:retraction, and a CSV file in a GitLab repository, refreshed once per working day.
If your reading already sits in Zotero, Zotero runs the retraction check for you. Zotero has flagged retracted items since version 5.0.67, announced on 14 June 2019, using Retraction Watch data: a flag in the items list, a warning with the retraction details at the top of the item pane, and a warning before you cite a flagged item through the word processor plugin. The catch is identifier coverage. Zotero’s announcement says the feature "is limited to items with a DOI or PMID", which it put at "about 3/4 of Retraction Watch data", so a saved item with neither identifier may not be flagged.
Litrevu’s reference checker reads the same Retraction Watch data through Crossref, alongside OpenAlex’s retraction flag, and covers references you never saved in a reference manager. The Litrevu guide on how to check whether a paper has been retracted walks through the manual route. Citing a retracted paper counts as the author’s error: the ICMJE Recommendations say "Authors are responsible for checking that none of the references cite retracted articles except in the context of referring to the retraction."
What do universities expect you to have checked?
Universities expect students to have verified every reference themselves, and some spell out the method. The University of Pretoria’s Artificial Intelligence Guidelines in South Africa warn that "AI frequently generates plausible-looking citations to non-existent papers. Never cite a source without verifying its existence and relevance", and instruct: "For every reference AI provides, search UP Library Catalogue, Google Scholar, or Web of Science using the complete citation (author, year, title)."
In the United Kingdom, the University of Oxford’s guidance on safe and responsible use of GenAI is shorter and broader. Oxford warns that "all tools can generate outputs that contain inaccuracies and fabrications", and that "You should always cross-check AI generated outputs against established sources to verify accuracy and identify erroneous information."
Rules differ by institution and by country, so the version that matters for a submission is your own faculty’s. The safest working assumption is the stricter of the Pretoria and Oxford positions: verify the existence of every reference, and check the detail of every reference against the source document.
Can an AI research tool check its own citations?
Partly, and the answer depends on where an AI tool gets its citations from. A tool that searches a real index returns records that exist, because each record came out of a corpus rather than out of a model’s memory, so the existence problem mostly disappears; what remains is the sentence-level question of whether the claim written next to a citation is the claim the paper makes.
Litrevu is an AI assistant for writing backed by sources. It reads every paper, report and PDF you add, answers your questions with a citation to the exact passage behind each claim, compares sources side by side, and checks your own draft against them. Each Litrevu citation opens to the highlighted passage on the page of the PDF it came from, and a pasted paragraph comes back with each claim marked Supported, Partly supported, Contradicted or Not in your papers. Litrevu’s how it works page is plain about the limit: its check "confirms that a citation points to a passage in your papers, not that the passage fully supports the sentence", so the reading is still yours to do.
Litrevu’s job is narrower than searching the literature, and other tools beat Litrevu at their own part of the work. Scite, for one, tells you whether later papers have supported or contrasted a paper you are about to cite, which the Litrevu check has no view of.
| Tool | Strongest at |
|---|---|
| Elicit | Discovery and systematic-review work across a corpus of more than 138 million papers, with screening and data extraction for systematic reviews |
| Consensus | Fast evidence search: an AI-powered academic search engine across peer-reviewed papers, with a Consensus Meter showing how many studies support or contradict a yes-or-no claim |
| Scite | Citation context: Smart Citations classify citing statements as supporting, contrasting or mentioning, so you can see how later work received a paper |
| Research Rabbit | Visual discovery: a citation-graph tool that starts from one paper and expands through related works, authors and topics, with built-in visualisations |
| Litrevu | Questions across papers you already chose, each answer opening to its passage, and checking your own paragraph against those papers |
Each row follows what that tool’s own pages say it is for. Plans change often, so this table carries no plan details; the Litrevu guide to the cheapest literature review software for postgraduate students covers the cost side.
What does a checked reference list look like?
A checked reference list is one where every entry carries a status, so you can see at a glance which entries need you to open the paper. Litrevu’s reference checker runs four checks per entry, on up to 30 entries per check: whether the work exists, whether the year, first author, volume, first page and DOI agree with the record, whether the DOI belongs to that paper, and whether a retraction or other notice is on record.
The block below is a format illustration, not a real result. It uses the reference checker’s own status labels, the placeholders stand in for whatever your list contains, and no author, title, year or finding in it is real.
FORMAT ILLUSTRATION ONLY, NOT A REAL RESULT
(Author, Year) Found and matches. The record agrees with your entry
(Author, Year) Found, but details differ. The record gives a different year
(Author, Year) The DOI belongs to a different paper. Your DOI is real but registered to another work
(Author, Year) No registration agency knows this DOI. A typo or a very new DOI also causes this
(Author, Year) We could not find this. Search the library catalogue by hand
(Author, Year) Found, but retracted. A retraction notice is on record for this paper
The checker says "We could not find this" and never "fake", because books, theses, reports and some regional journals are missing from these indexes, and a real source should not be called invented because an index lacks it. A "Found and matches" status says the paper exists and the details line up, and nothing more: the sentence written next to that reference is unverified until you read the paper, which is the check the Litrevu guide on how to spot fake AI citations before you submit sets out by hand.
How do you check a 60-item reference list?
Check a 60-item reference list in two passes: an automated pass for existence, details and retractions, then a manual pass on everything the automated pass flags and on every claim. Crossref’s Simple Text Query takes all 60 references in one submission, since its stated limit is 1,000 per submission, and Litrevu’s reference checker takes the list in two runs at its limit of 30 references per check.
The manual pass is where the work is. A flagged reference gets searched the way the University of Pretoria’s Artificial Intelligence Guidelines describe, by its complete citation (author, year and title), in a library catalogue, Google Scholar or Web of Science, one reference at a time. Then every sentence in the review that carries a citation needs the cited paper open beside it at least once, which is what the ICMJE Recommendations mean by being able to "attest that the references cited support the associated statement". Formatting the verified entries is a separate job again, covered in the Litrevu guides to Harvard referencing and APA citation format.
Frequently asked questions
What is the best tool to check citations in a research paper?
The best free combination is two tools: Litrevu's In-Text Citation Checker to match every in-text citation to the reference list, then a reference checker such as Crossref's Simple Text Query or Litrevu's reference checker to confirm each entry exists and has not been retracted. No tool confirms that a paper supports the sentence citing it; that check means opening the paper.
Which platforms provide reliable citation checking?
Crossref, OpenAlex and Zotero are reliable for what each covers because they read published records: Crossref matches references to registered DOIs and carries the Retraction Watch database, OpenAlex is an open catalogue of about half a billion works, and Zotero flags retracted items with a DOI or PMID. Litrevu's reference checker combines Crossref, OpenAlex and CORE with fixed rules and no AI model deciding the verdict.
Is there a free tool that checks if my references are real?
Yes. Crossref's Simple Text Query accepts up to 1,000 pasted references and returns the DOI matched to each, and Litrevu's reference checker takes up to 30 references per check with no sign-up and flags retracted papers. OpenAlex's data is free under CC0 and works well for checking a single title or DOI.
Is there a free alternative to referencechecker.org?
Yes. For checking whether references exist and have been retracted, Crossref's Simple Text Query and Litrevu's reference checker are free, and Litrevu's In-Text Citation Checker matches citations to the reference list for free, with no account.
What does a reference quality check cover?
A reference quality check covers four things: whether each work exists, whether its recorded details (author, year, title, volume, pages, DOI) match the published record, whether its DOI belongs to that work, and whether it has been retracted or carries an expression of concern. A fuller check adds whether every in-text citation has an entry and every entry is cited, and whether each paper supports the sentence that cites it.
Can a reference checker tell me if the paper actually supports my sentence?
No. A reference checker confirms that a work exists and that the bibliographic details match a record, which is a different question from whether the paper's findings support the claim you wrote. The ICMJE Recommendations ask authors to "attest that the references cited support the associated statement", and the only way to do that is to open the paper and read the relevant passage.
What should I do if a reference will not verify?
Search for the reference once by its complete citation, author, year and title, in a library catalogue or Google Scholar, because books, theses and conference papers often have no DOI and fail an automated check while being perfectly real. If a careful manual search still finds nothing, remove the reference and the claim resting on it, then find a source you can read for that claim.
Whose problem is it if a reference turns out to be fabricated?
The author's. University guidance in South Africa and the United Kingdom puts the check on the student, not on the tool: the University of Pretoria's Artificial Intelligence Guidelines say "Never cite a source without verifying its existence and relevance", and the University of Oxford tells students to "always cross-check AI generated outputs against established sources". For journal submission, the ICMJE Recommendations put the duty on the author, including "checking that none of the references cite retracted articles".
Ask Your Own Papers
Litrevu reads the papers you have already gathered and answers your questions about them, with every citation opening to the passage in your own library it rests on, so the sentence level check that no reference checker can do for you starts one click from the claim. Paste a paragraph you wrote to see which of its claims your papers back. Litrevu is free to start, and no credit card is required.
Start with your own papers