Repair a damaged PDF
A free PDF repair tool that runs in your browser, so you never upload the damaged file anywhere. It reopens the document with a reader that rebuilds a broken cross-reference table, writes a clean copy, and reports what was wrong and what came back.
Opens a damaged PDF with a tolerant reader, writes a clean copy, and reports what was actually wrong.
In detail
A PDF is a pile of objects plus a table saying which byte each one starts at. Edit the file with software that does not update that table, lose the end of a download, or let a sync client write half a copy, and every offset after the damage points somewhere wrong. Most readers stop there and call the file corrupt.
This tool opens it with pdf.js, which scans the whole file for objects when the table disagrees with it, then writes a new document with a correct table and trailer. It reports what it found first: the page count recovered, whether the index was stale, which pages failed. If the file already opens cleanly it says so instead of pretending to have fixed something.
Repair a damaged PDF at a glance
- Input
- One PDF, however damaged, including a file other readers refuse to open.
- Output
- One rebuilt PDF with a fresh cross-reference table and trailer, plus a written diagnosis on screen.
- What it does
- Reports what is wrong before it changes anything
- Rebuilds a damaged cross-reference table and trailer
- Keeps text, fonts and image bytes when it copies pages
- Names the pages it could not recover instead of dropping them quietly
- Counts junk in front of the header and bytes past the end of the file
- Redraws pages as images as a last resort, and says what that costs
- Detects a password-protected file rather than guessing at it
- Who it is for
- Anyone holding a PDF a reader calls corrupt: a download that stopped, a file off a failing drive, or a document an editor saved without updating its index.
- Limits
- Content that was truncated away or overwritten is not in the bytes, so nothing can bring it back.
- A password-protected PDF cannot be read or rewritten here.
- Redrawing pages as images keeps the look of each page and loses selectable text, links and form fields.
- A very large document is limited by the memory your browser gives one tab.
- Privacy
- The file is read and rewritten in this tab. Nothing about it is sent anywhere, including the file itself.
How to repair a damaged PDF file
- Drop the damaged PDF in Drop the file on the page or browse for it. It is read in this tab, so the diagnosis appears without the file being sent anywhere, however broken it is.
- Read the diagnosis The verdict line says whether anything is wrong at all. Under it every finding is listed most serious first: a stale cross-reference pointer, a missing end-of-file marker, junk in front of the header, pages that could not be read.
- Pick a rebuild method Keep the pages copies every page object into a clean document, so text stays text. Redraw as images is the last resort for a page tree the writer cannot parse, and it offers a redraw resolution from 110 to 300 DPI.
- Check the page list Every page is listed with the size it will come out at and whether it could be read. A page marked as unreadable is left out of the rebuilt file rather than written in empty.
- Rebuild and save Name the output and press Rebuild and save. A new PDF is written from scratch with a fresh cross-reference table and trailer, and the message afterwards says how many pages made it in.
Repair a damaged PDF: common questions
Can this recover a PDF that will not open at all?
Sometimes. The file is opened with pdf.js, which rebuilds a broken cross-reference table by scanning the whole document for objects, so a file other readers refuse often opens here and can be written back out clean. If the damage reaches the catalog itself, or the pages were truncated away, nothing opens and the tool says exactly that rather than saving an empty file.
Will the repaired file keep its text?
Yes, with the default method. Keep the pages copies each page object into the new document, so text stays selectable, fonts stay embedded and images keep the bytes they arrived with. Only the Redraw as images fallback loses that, and it warns you before you use it.
Why does it say nothing needs repairing?
Because the structure is intact: the header is at the front of the file, the cross-reference pointer resolves, the end-of-file marker is there and every page reads. A reader refusing a file like that is objecting to something else, such as a font it cannot load or a feature it does not support. Rewriting it is still offered, but it will not change anything.
Can it remove a password from a PDF?
No. An encrypted PDF has every stream scrambled, and without the password there is nothing to read or copy. Encryption is detected and reported instead of being worked around, because a rewrite of unreadable streams would just be a file full of noise. Open the document in a reader you have the password for, save an unprotected copy, and repair that.
Does the file get uploaded to be repaired?
No. The reader, the writer and the rebuild all run in this tab, so the bytes never leave your machine. That also means the size a document can be is set by the memory your browser gives one tab rather than by an upload limit, and a scan that has to look at every object in a very large file takes a moment.