What kind of damage can be fixed?

Wrong cross-reference offsets, a missing trailer or startxref, junk before the header, wrong stream lengths, a broken page tree and a lost document catalog are structural problems that can usually be fixed completely without losing content. For truncated files, pages before the cut are normally recovered, the page where the cut happened may be partial, and anything after it is simply no longer in the file.

What’s the difference between structural repair and re-rendering?

Structural repair keeps the original text, fonts and vector graphics, so text stays selectable and searchable — it’s always tried first. Re-rendering draws every page with PDF.js and builds a new PDF from the images: it looks the same, but text becomes part of the picture. It’s used automatically only when structural repair fails.

Can encrypted PDFs be repaired?

Structural repair works and the result keeps the original encryption, so it still asks for the password. If you know the password you can unlock the repaired file with Unlock PDF afterwards. Re-rendering needs to open the file, so it doesn’t work for PDFs with an open password.

Is my file uploaded? Does repair change the content?

Nothing is uploaded; it all happens in your browser. Structural repair rewrites only the file structure (cross-reference table, catalog, page tree) and leaves page content as it was. Data lost to truncation can’t be conjured back, and the report lists any incomplete objects that had to be dropped.

🩹

🩹Repair PDF Online - Fix Damaged, Corrupt or Truncated PDFs

A PDF that won’t open, complains that it is damaged or that the cross-reference table can’t be read, or was cut short by an interrupted download? The tool starts with a check-up: is the %PDF header pushed back by junk bytes, is the %%EOF marker missing, does startxref point at a real cross-reference table, are the object offsets in that table correct, are the trailer and document catalog (/Root) still there, and does each stream’s /Length match its data? Every problem found is listed. Repair first reads all objects with a lenient parser that doesn’t rely on the broken offset table. If that fails, or the file is truncated, it scans for every ‘N 0 obj … endobj’ itself and rebuilds a clean file — keeping the last complete version of each object number and decompressing as much of a truncated stream as possible, so even half a page of text can survive. It then fixes the document catalog and page tree: a missing catalog is located again or recreated, a broken page tree is rebuilt from the page objects found, missing page sizes and resources are filled in, and references to objects that no longer exist are removed. The result is checked page by page with PDF.js; if the structure really can’t be saved, PDF.js — the most forgiving parser — renders every page it can display into a new image-based PDF. A repair report shows what was wrong, what was fixed and how many pages were recovered. Everything runs in your browser; files are never uploaded.

Free · No sign-up · Runs in your browserLast updated: 2026-10-26

Loading tool...

How to use Repair PDF

1Upload a PDF that won’t open or shows errors; it is diagnosed and the problems are listed
2Choose structural repair (keeps text, falls back to re-rendering) or re-render straight away
3Click Repair & download, read the repair report and save the fixed PDF

FAQ

What kind of damage can be fixed?: Wrong cross-reference offsets, a missing trailer or startxref, junk before the header, wrong stream lengths, a broken page tree and a lost document catalog are structural problems that can usually be fixed completely without losing content. For truncated files, pages before the cut are normally recovered, the page where the cut happened may be partial, and anything after it is simply no longer in the file.
What’s the difference between structural repair and re-rendering?: Structural repair keeps the original text, fonts and vector graphics, so text stays selectable and searchable — it’s always tried first. Re-rendering draws every page with PDF.js and builds a new PDF from the images: it looks the same, but text becomes part of the picture. It’s used automatically only when structural repair fails.
Can encrypted PDFs be repaired?: Structural repair works and the result keeps the original encryption, so it still asks for the password. If you know the password you can unlock the repaired file with Unlock PDF afterwards. Re-rendering needs to open the file, so it doesn’t work for PDFs with an open password.
Is my file uploaded? Does repair change the content?: Nothing is uploaded; it all happens in your browser. Structural repair rewrites only the file structure (cross-reference table, catalog, page tree) and leaves page content as it was. Data lost to truncation can’t be conjured back, and the report lists any incomplete objects that had to be dropped.