Skip to main content

Repair PDF

Repair PDF

Rebuild a broken PDF structure, and see exactly what was wrong and what changed.

Drop damaged or corrupted PDF file(s) here

Fixes broken headers, cross-reference tables and page trees • Lists anything it cannot recover

Max 100MB per fileUp to 50 files

How to Repair and Recover Damaged PDF Documents

How to Use Repair PDF

  1. 1

    Upload Damaged PDF

    Select the broken or unopenable PDF file.

  2. 2

    Choose Options

    Decide whether lost pages get a placeholder, whether blank pages are removed, and whether the document information is cleared. There is no mode to pick: the file is checked and fixed in place.

  3. 3

    Read the Report

    The result lists what was actually found wrong, what was changed, and anything worth knowing about the output.

  4. 4

    Download Repaired PDF

    Save your recovered document ready for viewing and editing.

What Repair PDF Can Do

  • Repairs In Place

    Writes a new cross-reference table and rebuilds the page tree only when it is broken. Bookmarks, forms, attachments and metadata stay as they were.

  • Header Junk Removal

    Finds the %PDF- signature within the first 1 MB and drops anything in front of it, or adds a header that is missing altogether. A common cause of "not a PDF" errors.

  • Honest Diagnostics

    Reports the problems it actually detected, separately from the changes it made, and lists anything it could not recover by page number. A healthy file is told it is healthy.

  • Last-Resort Page Rescue

    When the structure cannot be parsed at all, every page is re-rendered as an image inside a new document. Readable, but the text layer is gone — and the report says so.

  • Private & Secure

    Salvage sensitive corporate documents securely without uploading them to cloud servers.

Tools & Settings Reference

Keep a placeholder for lost pages
A page that cannot be recovered is replaced by a page that says so, so the pages after it keep their numbers. Turn it off to leave lost pages out; either way they are listed by number.
Remove blank pages
Renders each page and drops the ones that come out visually empty, plus pages whose page box has collapsed. The two are counted separately in the report.
Clear document info
Deletes the title, author and every other document property, and the XMP metadata. Useful when that block is the damaged part. Left off, they stay exactly as they are.
Compress output streams
Packs objects into compressed streams for a smaller file. Turn it off to write a larger file that viewers older than PDF 1.5 can still open.

Frequently Asked Questions

Stray bytes in front of the %PDF- header or a missing header, a cross-reference table or stream whose offsets no longer match the objects, a missing or wrong startxref pointer or %%EOF marker, out-of-range numbers that make a strict parser give up, streams whose stated length is wrong, a broken page tree or document catalog, pages that lost their type, and pages that have collapsed to a zero-sized box. A file that was cut short keeps every complete page. What cannot be recovered is content that is no longer in the file: those pages are listed by number instead of disappearing without a word.

The tool falls back to re-rendering every page it can display and embedding those renders as images in a new document. That is genuinely the last resort: the pages will look right, but there is no selectable text, no working links and no form fields, and the file is usually larger than the original. The result screen flags this prominently rather than filing it under "repaired" — keep your original copy either way.

Then it says so. The report distinguishes "no damage found" from "repaired", and lists the damage it detected separately from the changes it made. A clean file is written out again with a fresh cross-reference table and nothing else changes, so you will not be shown a list of imaginary repairs. If the file is digitally signed, that re-save invalidates the signature; the tool warns you, and the original stays the signed record.

Not normally. The pages are repaired where they are, so text, fonts, images and vector art come out identical, and the document information and XMP metadata stay exactly as they were unless you ask for them to be cleared. The exceptions: a page that could not be recovered is replaced by a labelled placeholder, or left out if you turn placeholders off; a page whose content was lost comes out blank and is listed; and the raster fallback above rebuilds pages from images.

Related Tools