How to fix import errors

Import problems surface at one of two moments: when a file is uploaded, or when it is processed into references. The message tells you which — this article maps each one to its fix.

Files rejected at upload

These appear on the upload page, before anything is stored:

“Some files were rejected because they are not a supported format (.bib, .ris, .txt, .xml).” — The file’s extension isn’t one Littview accepts. Re-export from your database in BibTeX, RIS, PubMed, or EndNote XML (the formats article says which databases produce which). Don’t rename a CSV or Excel export — the content check at processing will catch it anyway.
“Exceeds the 10MB limit” — shown next to the individual file; the others in the batch still upload. Reference exports this large are rare — if you hit it, split the export in your database (for example by year range) and upload the parts.
“You can upload up to 20 files at once.” — The whole drop is rejected, nothing partial. Upload in batches of 20 or fewer.

Processing stops with an error

Processing imports libraries one at a time and stops at the first one it can’t handle, naming it:

“Unsupported file format for library: X” — The file uploaded fine but its content isn’t a recognized format. Littview reads the first line of the file, not the extension — so this usually means the export option was wrong (a CSV renamed .ris, an RIS chosen where the site actually produced plain text) or the file is empty. Delete that library, re-export correctly, upload, and process again.
A reference-limit message with numbers — The file would take the project past your plan’s reference cap, so it was refused whole; nothing from it was imported. Trim the export (tighter search, or split and import what fits) or raise the cap, then process again.

A stopped run is safe to rerun: libraries that already imported are skipped, and the one that failed retries cleanly once fixed.

Fewer references than the export contained

Entries the parser can’t read — usually hand-edited records or an export quirk, such as an unclosed brace in a BibTeX record — are skipped; the rest of the file imports normally. The tell is the count: the processing summary’s “references added” number falls short of what your export contained (most databases show the record count when you export). If a shortfall matters, re-export the file fresh from the database rather than hunting for the broken record — then delete the library (Owner only) and import the new file.

Fixing individual records after import

Imported but imperfect records don’t need a re-import:

DOI and abstract can be edited directly on the reference: open it on the References tab and use the inline editors (Owner and Assistants, on an unarchived project). This matters later — the abstract is what reviewers screen against.
Anything else (title, authors, year, journal) comes from the source file. The clean fix is to delete the affected library (deleting is Owner-only) and re-import a corrected export: deleting a library removes its references, so do this before screening starts — decisions made on those references would go with them.