Checking article coverChecking article cover

DOI Checker: Check That a DOI Works and Opens the Right Paper

To check whether a DOI is right, run three checks. First, the form: a DOI starts with "10.", and a slash separates its prefix from its suffix (DOI Foundation, 2025). Second, put doi.org/ in front of it and see whether it opens; if doi.org can't find it, you get an error page headed "DOI NOT FOUND." Third, read the page it opens and compare the title and authors with your reference. Don't skip the third check. When we changed the last digit of a real DOI from 8 to 9, doi.org opened a different published paper by another author. For a whole reference list, Crossref's Simple Text Query and PerfectRefs both take a pasted list.

We ran every test in this article on November 11, 2026: each DOI at doi.org, then the Crossref or DataCite record for each one that opened.

What does a DOI checker need to check?

CheckHow to run itWhat a failure tells you
1. FormStarts with 10., then the rest of the prefix, a slash, and the suffixWhat you copied is not a complete DOI
2. It opensdoi.org takes you to a landing pageNobody registered this DOI, or it contains a typo
3. Right paperThe landing page shows your title and authorsThe DOI belongs to another work

Check 1 is quick but limited. The suffix is "an arbitrary string" chosen by the registrant, and the DOI Handbook sets no limit on the length of a DOI or its parts (DOI Foundation, 2025). A wrong DOI can look exactly like a right one. What you should strip before checking is anything wrapped around the DOI: the "doi:" label is not part of the DOI (DOI Foundation, 2025), and Crossref names a period at the end of a DOI as one way a link gets "formatted incorrectly" (Crossref, 2024). For how each citation style writes a DOI, see DOI format in references.

How do I check one DOI by hand?

  1. Copy the DOI from your reference, starting at "10." and leaving out any label or link in front.
  2. Open https://doi.org/ followed by the DOI, or paste it into the single box on the doi.org home page, which asks for "a known DOI name" (DOI Foundation, n.d.-a).
  3. On the landing page, compare the title, first author, year, and journal with your reference.
  4. If the landing page won't load or doesn't show those details, read the registration record instead. For a Crossref DOI, add the DOI to the end of api.crossref.org/works/, as in https://api.crossref.org/works/10.7759/cureus.39239 for the Lingas paper below. In our tests it returned the title, authors, journal, and publication date.

If you don't have a DOI yet and need to find one from a title, use our DOI lookup guide.

Can a DOI open and still be wrong?

Yes. We took the DOI of a 2023 Cureus study and changed its last digit, one way and then the other:

DOI triedWhat doi.org opened
10.7759/cureus.39238 (the real DOI)Bhattacharyya et al. (2023), "High rates of fabricated and inaccurate references in ChatGPT-generated medical content"
10.7759/cureus.39239 (last digit 8 changed to 9)Lingas (2023), "Hematological abnormalities in cirrhosis: A narrative review"
10.7759/cureus.39237 (last digit 8 changed to 7)A third paper, a systematic review of rituximab in CANOMAD

Both altered DOIs are registered and open without an error, so checks 1 and 2 pass them. Only check 3 catches the problem: the title on the landing page is not the title in your reference.

Not every slip lands on another paper. We also changed single digits in a Scientific Reports DOI, 10.1038/s41598-023-41032-5, five different ways, and doi.org found none of the five. Yet changing two characters, to 41033-4, opened a paper on OLED microdisplays, the example in our reference list checklist. Because the registrant chooses the suffix (DOI Foundation, 2025), you can't tell in advance whether a typo will hit another paper or nothing at all. So open every DOI and read what it opens.

The same failure turns up in references written by chatbots. In a 2024 study of references generated by a large language model, DOIs that "did not exist or were matched to a different source" were both counted as hallucinations (Mugaanyi et al., 2024). The Cureus paper we altered is itself a study of chatbot references: ChatGPT-3.5 wrote 30 short medical papers with 115 references, and 47% of those references were fabricated while 46% were authentic but inaccurate (Bhattacharyya et al., 2023). Hallucinated citations explains why chatbots produce them.

Why does a working DOI come back "not found" in Crossref?

Crossref is one DOI registration agency among several. The DOI Foundation lists 12, including Crossref, DataCite, JaLC, and mEDRA (DOI Foundation, n.d.-b). DataCite "provides DOIs for research data and all other research outputs" (DOI Foundation, n.d.-b). Here is what we got for the DOI of the arXiv preprint "Attention Is All You Need":

Where we lookedResult
https://doi.org/10.48550/arXiv.1706.03762Opened the arXiv abstract page, which labels it "arXiv-issued DOI via DataCite"
https://doi.org/doiRA/10.48550/arXiv.1706.03762"RA": "DataCite"
https://api.crossref.org/works/10.48550/arXiv.1706.03762404 Not Found
https://api.datacite.org/dois/10.48550/arXiv.1706.03762The record: title, 8 authors, publisher arXiv, year 2017

The DOI works, but Crossref has no record of it. A checker that looks DOIs up only in Crossref would report it as missing. Before you treat a DOI as wrong, ask doi.org who registered it: its Which RA? service "returns the name of the DOI Registration Agency (RA) responsible for a specific DOI" (DOI Foundation, 2020). Then look the DOI up in that agency's records.

Bulk DOI check: how to check a whole reference list

The doi.org box asks for one DOI name, which is slow going for a long list. Crossref's Simple Text Query works on a list. You paste up to 1,000 references per submission, separated by blank lines, and it "attempts to find exactly one DOI for each reference" (Crossref, 2020). Crossref sets no limit on how often you use it. Two things to know before you rely on it:

  • It finds no match when "the metadata in either the reference or record registered with us is not sufficient to make a single match" (Crossref, 2020).
  • The records in that sentence are the ones registered with Crossref. The arXiv DOI above is registered with DataCite, so a blank result for a DOI like that does not mean the DOI is wrong.

Its page describes finding a DOI for each reference, not checking the DOI you already wrote. To use it as a checker, put the DOI it returns next to yours and open every pair that differs.

In PerfectRefs, you paste the reference list; you don't upload the paper. The Style check compares each field with the database record, the DOI field included. When the search finds the paper you cited, it shows the record's DOI with a link to the source, and you can accept it with one click. It uses 2 premium credits per reference, and new accounts get 10 premium credits. The Existence check comes with a free daily allowance. It looks up each reference in Crossref, OpenAlex, and other scholarly databases and flags the ones it can't find. When a cited DOI opens a paper with a different title and the search finds no better match, it flags that reference for you to check. doi.org checks one DOI at a time; both checks run on the whole list at once.

FAQ

Is my DOI wrong if only its capital letters differ?

No. The DOI Handbook makes DOIs "case-insensitive only when testing for equivalence and only with respect to the Basic Latin Unicode block" (DOI Foundation, 2025). So when you compare your DOI with one a database returns, A and a count as the same, but every other character must match exactly.

Does every article have a DOI?

No. APA's rules cover journal articles without one: if the article came from an academic research database, the reference ends after the page range (American Psychological Association, 2025). To look for a DOI, see the DOI lookup guide.

If a DOI doesn't open, is the reference fake?

Not necessarily. Crossref's list of reasons a DOI fails includes a DOI that was published but not registered, a published DOI that doesn't match the registered one, a badly formatted link, and typing or copy-and-paste errors (Crossref, 2024). A DOI that hasn't been registered yet gets the same "DOI NOT FOUND" page at doi.org, and a DOI just submitted for registration can wait in Crossref's queue "between several minutes and several hours" (Crossref, 2025). Search for the title before you decide the reference doesn't exist.

References

  1. American Psychological Association. (2025, August). Journal article references. APA Style. https://apastyle.apa.org/style-grammar-guidelines/references/examples/journal-article-references
  2. Bhattacharyya, M., Miller, V. M., Bhattacharyya, D., & Miller, L. E. (2023). High rates of fabricated and inaccurate references in ChatGPT-generated medical content. Cureus, 15(5), Article e39238. https://doi.org/10.7759/cureus.39238
  3. Crossref. (2020, October 6). Simple Text Query. https://www.crossref.org/documentation/retrieve-metadata/simple-text-query/
  4. Crossref. (2024, July 19). DOI error report. https://www.crossref.org/documentation/reports/doi-error-report/
  5. Crossref. (2025, December 10). Verify your registration. https://www.crossref.org/documentation/register-maintain-records/verify-your-registration/
  6. DOI Foundation. (n.d.-a). Digital Object Identifier (DOI) system. Retrieved November 11, 2026, from https://www.doi.org/
  7. DOI Foundation. (n.d.-b). Existing registration agencies. Retrieved November 11, 2026, from https://www.doi.org/the-community/existing-registration-agencies/
  8. DOI Foundation. (2020, July 4). DOI resolution documentation. https://www.doi.org/the-identifier/resources/factsheets/doi-resolution-documentation
  9. DOI Foundation. (2025). DOI handbook. https://www.doi.org/doi-handbook/DOIHandbook_2025.pdf
  10. Lingas, E. C. (2023). Hematological abnormalities in cirrhosis: A narrative review. Cureus, 15(5), Article e39239. https://doi.org/10.7759/cureus.39239
  11. Mugaanyi, J., Cai, L., Cheng, S., Lu, C., & Huang, J. (2024). Evaluation of large language model performance and reliability for citations and references in scholarly writing: Cross-disciplinary study. Journal of Medical Internet Research, 26, Article e52935. https://doi.org/10.2196/52935

Original article by PerfectRefs. To republish it, credit PerfectRefs and link to this page. Do not edit it or use it commercially.