A DOI should resolve and its registered metadata should agree with the citation. Resolution alone does not prove that the citation is accurate, that the article is unchanged, or that the research is trustworthy.
Use a three-part verification
- Normalise and resolve the DOI. Remove prefixes such as “doi:” and test the identifier through the DOI resolver.
- Retrieve registered metadata. Check title, authors, publication year, container title and publisher through the relevant registration agency record.
- Compare the record with the reference. Record exact matches, explain material differences and avoid silently selecting a near match.
Interpret the result cautiously
A DOI is designed as a persistent identifier for an object. Its resolver can redirect to the current registered location, while registration-agency services may expose descriptive metadata. A working DOI therefore supports identity and discoverability, but it does not independently validate the study design, peer review, journal quality or accuracy of every citation field.
Worked use case
A reference contains the right DOI but the title and year differ from the registered record. Keep the DOI result, label the citation as a metadata mismatch and inspect the publisher record before correcting the reference. Do not describe the work as fabricated merely because one field differs.
Evidence boundary
Crossref does not register every DOI in existence, and metadata quality depends on deposits from registrants. A record not found in one registry may belong to another DOI registration agency. Absence from one lookup is therefore not proof that the identifier is false.
Primary sources
- DOI Foundation. DOI Handbook. Retrieved 4 August 2026.
- Crossref. REST API documentation. Retrieved 4 August 2026.