Repair PDF
Attempt to fix corrupted or damaged PDF files.
Drop the broken PDF here
Even damaged files — we'll try to recover them
Choose File
Repairing...
Attempt to fix corrupted or damaged PDF files.
Even damaged files — we'll try to recover them
Most PDF damage is structural rather than a loss of content. A PDF keeps an internal index of where every object sits in the file, and that index is easy to break: an interrupted download, a file copied from failing storage, a truncated email attachment, or software that wrote the trailer incorrectly. The content is all still there; the map to it is wrong.
The first method parses what remains, rebuilds the cross-reference table from scratch, and writes a clean, correctly indexed document. This is fully lossless. Text stays selectable and searchable with its original fonts, images keep their exact pixels, vector art stays vector, and form fields keep working. When Method 1 succeeds, you have your document back exactly as it was — and it succeeds for the large majority of files people bring here.
When the structure is damaged past rebuilding, the second method changes tactics: it renders each page that can still be drawn and assembles those into a fresh document. It works page by page, so a file with a corrupt region in the middle still returns everything either side of it — pages that cannot be recovered are reported and skipped rather than failing the whole job.
What comes back is the visual document: every page, in order, looking as it should. Because the pages are rebuilt from what could be rendered, the text layer does not carry over — and there is a good route from there. Run the result through OCR and it reads the pages and writes a fresh text layer, restoring search, selection and copy across the whole document. Two steps, and a file that would not open at all becomes a fully searchable PDF.
Before assuming damage, rule out the ordinary explanations: a file that prompts for a password is encrypted rather than broken, and Protect PDF removes the password when you know it. A file that opens but shows blank pages is usually a scan with no text layer, which OCR resolves.
The message you get tells you what you are holding.
Repaired by Method 1 — the best result. Your document, complete and intact, with its text layer, fonts and image quality unchanged. Nothing further is needed.
Repaired by Method 2, with a page count — the pages were recovered visually. Check that the count matches what you expect, then run OCR to restore the searchable text layer.
Individual pages reported as skipped — those regions of the file were unreadable, and everything else came through. Partial recovery of an important document is usually far better than none, and the pages you have are complete.
Whatever the outcome, keep the damaged original until you have checked the repaired file. It costs nothing, and a different copy of the same file — from an email thread, a backup, a colleague — sometimes repairs more cleanly.
A recovered document is worth a few minutes of finishing. Run OCR if Method 2 produced the result, so the text is searchable again. Check the properties with Edit Metadata, since a rebuilt file often loses the title and author. If pages are missing from the middle, Insert Pages puts a recovered page back at the right position from another copy. And once you have a good version, Compress PDF keeps the archived copy compact.
Structural damage — a broken index, a truncated trailer, a malformed revision — repairs cleanly and losslessly, and that covers most of what people encounter. Damage to the content regions themselves is handled page by page, so a file with one corrupt area still returns every page either side of it. A file whose bytes are almost entirely lost has little left to recover, which is true of any tool.
Repair reconstructs what is present in the file you have; it does not restore content that was removed and saved over. If an earlier version exists in a backup, an email thread or a cloud version history, that copy is the better starting point.
Method 2 rebuilds pages from what could be rendered, so run the result through OCR to write a fresh text layer. That restores search, selection and copy across the whole document — and it is a quick step. Method 1 keeps the original text layer intact and needs nothing further.
Often it drops, because a rebuild discards the accumulated revision history and orphaned objects that damaged files tend to carry. Method 2 output size depends on the page content. Either way, Compress PDF handles the final size.
No — that is encryption working as intended. Protect PDF removes the password when you supply it. Repair is for files that fail to open or render at all.
Blank-looking pages that open normally usually mean a scanned document with no text layer, or content drawn in a way an older reader cannot handle. Try OCR first for the scan case. If the file genuinely will not render, repair is the right tool.
Often, yes — that is a common reason people arrive here. Many readers are strict about specification compliance and reject files that are merely unusual rather than broken. A structure rebuild normalises them into something every reader accepts.
No. Both recovery methods run entirely inside this browser tab, and the repaired document goes straight to your downloads. Nothing is transmitted or stored, so a confidential file that has been damaged can be recovered without sending it to anyone.
Privacy: All repair logic (parsing, rendering, rebuilding) runs in your browser with PDF.js and pdf-lib. Damaged financial or medical PDFs never leave your device.