Fix a damaged or half-downloaded PDF and see exactly which pages could be recovered
Processed in your browser. Nothing is uploaded. Checked against sources on 22 Oct 2026.Repair PDF reads a damaged PDF in your browser, rebuilds its cross-reference table and trailer, and copies every page that still opens into a clean file. Each page is checked by a second engine and listed as copied, redrawn or lost. Files up to 100 MB work, and nothing is uploaded.
Press Select PDF or drop the damaged file on the page; it is opened on your device and never uploaded.
The file is checked and rebuilt at once: every page that can still be read is copied into a clean PDF.
Read the result: which pages kept their text, which were redrawn as pictures and which were lost.
Press Download to save the repaired PDF, and Report for a text list of what was wrong.
A PDF that "failed to load" or "is damaged" is rarely damaged everywhere. Most often a small part is broken: the table at the end of the file that tells a reader where each object starts, the trailer that says where the document begins, or the last few kilobytes lost when a download stopped. Readers give up because they cannot find the pages, even though the pages are still there.
What the repair does. The page reads the file the way the PDF specification lays it out: header, objects, cross-reference table, trailer. It finds every object that is still complete, drops the ones that were cut off, and writes a fresh table and trailer. When the page list itself is gone, it is rebuilt from the pages found. Then every page is copied into a new PDF with pdf-lib, and the new file is opened with pdf.js to check that each page really reads.
| Outcome | What it means |
|---|---|
| Copied | the page came through with its text, pictures and links |
| Drawn as a picture | its structure was broken; it looks the same but its text is no longer selectable |
| Copied, nothing to draw | the page was blank, or its content was lost |
| Lost | nothing of it is left in the file |
Honest about limits. Nothing is invented. A page that is not in the file is reported as lost, with its number, and the downloadable report lists every outcome. Bookmarks, attachments and form fields are not carried over. A password-protected file has to be unlocked first.
Tested on broken files. The repair was checked on files written byte by byte and then broken on purpose: wrong offsets, a missing trailer, a lost page list, extra bytes in front, a download cut in the middle of a page.
On your device. The file never leaves your browser.
Drop the file on Repair PDF. The page reads what is left of it in your browser, rebuilds the parts readers use to find pages, and copies every page that still opens into a clean PDF. The result lists each page as copied, drawn as a picture or lost, and Download saves the repaired copy.
Partly. A download that stopped keeps the pages written before the cut, and those come back with their text. Pages stored after the cut were never in the file, so no tool can bring them back; Repair PDF names them as lost instead of hiding the gap. Downloading the file again is the only way to get them.
A page whose internal structure is broken cannot be copied as it is, but pdf.js can often still draw it. Such a page is redrawn as a picture at print-friendly resolution, so it looks the same, but its text can no longer be selected or searched. The result says which pages this happened to.
Content that is not in the file cannot be recovered: the missing end of a truncated download, or pages overwritten with other data. Password-protected files must be unlocked first with Unlock PDF. Bookmarks, attachments and fillable form behaviour are not carried over; each page's text, pictures and links are.
No. The file is read and rebuilt by pdf-lib and pdf.js inside your own browser, and the repaired copy is made on your device. Nothing is sent to a server, so a confidential contract or statement stays private. Closing the tab clears the file from memory.
Every copied page is opened again with a second PDF engine before the result is shown, and a page only counts as recovered if that engine can read it. The list on screen and the downloadable report give the page numbers for each outcome, so you can compare them with the original.
Rates and rules change; check the source before you rely on a figure.
Still stuck? Ask about Repair PDF and anybody who knows can answer. No account needed.
Same job, other tools: HEIC to PDF – Convert iPhone HEIC photos straight into a PDF · JPG to PDF – Convert JPG photos into a single PDF, in the order you choose · Compress PDF to 500 KB – Fit a multi-page scan under 500 KB with its text still readable · Compress PDF to 50 KB – Get a one-page scan under 50 KB for strict document uploads · Resume Word to PDF – Send your resume as a PDF so the formatting holds · Extract PDF Pages – Pick the pages you need and save them as a new PDF.