1. Home
  2. PDF
  3. Repair PDF

Repair PDF

Diagnose a damaged PDF and repair it step by step.

Runs in your browser · No upload

How to use

  1. Drop a PDF that will not open (or click to choose it). As soon as it loads, a diagnosis list appears.
  2. Click Try to repair. The steps below run automatically, and later steps are skipped once one works.
  3. On success you get name_repaired.pdf, a table of what worked and failed, and the page count (estimated before → after).
  4. If every step fails, you see “Could not repair this file” with likely causes.

Repair steps

Step What it does Text and images
1. Diagnose Checks the header position, the end of file (%%EOF), the startxref value, and whether it is HTML or ZIP Unchanged
2. Rewrite structure Reads the file with a forgiving parser and saves a fresh structure; junk at the start is cut off Kept
3. Rebuild xref Scans the whole file for N G obj … endobj, builds a new cross-reference table and trailer, then repeats step 2 Kept
4. Pages as images Renders only the pages the browser’s PDF engine can open into a new PDF Text cannot be selected or searched

Each step counts as a success only after the saved file is opened again and its page count is checked. Pages that cannot be rendered in step 4 are left blank and listed as “pages N could not be recovered”. You can choose step 4’s resolution (150, 200 or 300 dpi, default 200) and JPEG quality (default 85%).

Common causes of damaged PDFs

  • Interrupted download: the end of the file is missing, so there is no %%EOF or page location table.
  • Damaged email attachment or chat transfer: extra bytes are added at the start or line endings are converted, so stored positions no longer match.
  • Closed while saving: the program quit mid-save and only half of the file was written.
  • Saved web page: a login or error page is downloaded with a .pdf name (shown as HTML in the diagnosis).

Example results

These are results from cutting a simple 5-page test PDF created with this tool (no object streams):

Portion kept Diagnosis Result
100% (intact) No problem 5 pages from the structure rewrite
70% End of file cut off 4 pages from the structure rewrite
40% End of file cut off 1 page from the structure rewrite
10% End of file cut off Cannot be repaired (no openable page)

Real files differ with their structure. This table shows values for a simple test file; in general, pages after the cut-off point cannot be brought back. If the page count dropped, download the original again and compare.

When repair is not possible

  • Files that barely downloaded (like the 10% case above, with no page information).
  • Files that are not PDFs, such as HTML, ZIP or empty files: the diagnosis says so and no repair is attempted.
  • Password-protected files: the structure can be fixed without the password, but reading page content, as in step 4, needs it. The original password still applies after a structure repair.
  • Files whose compressed data inside is itself damaged: the structure can be fixed while that page’s content may still look empty.

Good to know

  • A repaired file can differ in size from the original. To shrink it, use Compress PDF.
  • To remove a password you know or lift editing limits, see Unlock PDF. To get pages as pictures, use PDF to JPG.
  • Files up to 300 MB and 2,000 pages are supported.

Your file is processed only in your browser and is never sent to a server, so sensitive documents are safe to repair here.

Can't find the tool you need?

Tell us what you want to do. We build the most requested tools first.