How to repair a corrupted PDF and recover what is still inside it

A PDF that will not open feels like a total loss, but it usually is not. In the overwhelming majority of cases the page content is still sitting in the file, intact and readable — what has broken is the small piece of bookkeeping that tells a PDF reader where each page begins. Repairing the file means rebuilding that bookkeeping, not recreating the document.

This guide explains what actually breaks, how to tell a recoverable file from a genuinely destroyed one, and how to repair a PDF in your browser without installing anything.

What actually breaks in a corrupted PDF

A PDF is not a single continuous stream of text the way a plain file is. It is a collection of numbered objects — fonts, images, page descriptions, metadata — followed by a lookup table at the very end of the file called the cross-reference table, or xref. The xref records the exact byte offset of every object. When a reader opens a PDF, it jumps to the end, reads the xref, and uses it to find page one.

That design is fast, but it has a specific weakness: the index lives at the end and points backwards. If anything shifts the position of the objects, or the tail of the file never arrives, every offset in the xref becomes wrong at once. The reader looks for page one at byte 48,213, finds something that is not a page, and gives up. Nothing about the page itself has been damaged.

• Broken cross-reference table. The index is present but its offsets no longer match reality. This is the single most common form of PDF corruption and it is almost always repairable.
• Truncated file. An upload, download, or copy stopped early, so the last portion of the file — including the xref — never arrived. Recoverable up to the point where the file stops.
• Damaged object streams. Individual objects have been altered, usually by a program that wrote to the file incorrectly. Repairable page by page; damaged pages may lose formatting.
• Corrupted embedded fonts. The text is present but the font it references is unreadable, so pages render as blank or as boxes. Usually fixable by substituting a standard font.
• Invalid header. The file does not begin with %PDF-. Often means the file is not really a PDF, or that something has been prepended to it.

Common error messages, decoded

The message your reader shows is a reasonable guide to what went wrong. These are the ones that come up most often:

That last row matters more than it looks. A meaningful share of files people bring to a repair tool are not damaged — they are password-protected, and the reader's message is simply unclear. If you know the password, use Protect PDF to manage the file's encryption rather than trying to repair it.

What can and cannot be recovered

Setting expectations honestly is worth more than optimism here, because it tells you when to stop spending time on a file and go looking for the original instead.

• Nearly always recoverable: broken xref tables, files that a reader refuses to open but that are otherwise complete, PDFs that were saved by a program that crashed mid-write, and files that render blank because of font problems.
• Partially recoverable: truncated files — you get everything up to the byte where the file stops, and nothing after. A download that reached 80% typically yields roughly 80% of the pages, intact.
• Rarely recoverable: files where the object streams themselves have been overwritten with unrelated data, for example by a failing disk or a bad memory card. The index can be rebuilt but it will point at content that is genuinely gone.
• Not a repair problem at all: encrypted files, files that were never PDFs (a .pdf extension on a Word document, say), and 0-byte files. There is nothing inside a 0-byte file to recover.

Three things to try before repairing

Each of these takes under a minute and each of them solves a real category of problem that is not corruption:

1. Open it in a different reader. Readers vary enormously in how forgiving they are about malformed files. A PDF that Adobe Acrobat rejects outright will often open in Chrome or Firefox, which repair minor structural problems silently. If it opens in a browser, print it back to PDF and you have a clean copy.
2. Download it again. If the file arrived over a network, the simplest explanation is that the transfer failed. Re-download it, and compare the file size against the source before assuming the file itself is at fault.
3. Check that it is a PDF. Open it in any plain text editor and look at the first line. A real PDF begins with %PDF-1.4 or similar. If you see PK, it is a ZIP or an Office file with the wrong extension; if you see <!DOCTYPE html>, you downloaded an error page instead of the document.

Repair a PDF in three steps

If the file is genuinely damaged, repairing it with PdfDocShift takes three steps. There is no installation and no sign-up, and every uploaded file is permanently deleted from our servers after two hours.

Repairing scanned and image-heavy PDFs

Scanned documents corrupt differently from text documents, and they are worth treating as their own case. A scanned PDF is a container of large images, so a far greater share of the file is image data — which means truncation costs you whole pages rather than a broken index, and a partial file often ends mid-image, leaving a final page that renders half-drawn.

Two things follow from that. First, repair a scanned PDF before doing anything else to it; running compression or conversion on a damaged scan tends to bake the damage in permanently. Second, if the repaired file opens but the text is no longer selectable, the text layer did not survive — you can rebuild it by running the repaired file through OCR PDF, which reads the images and writes a fresh searchable layer underneath them.

If the recovered scan is unwieldy afterwards — repair sometimes produces a larger file than the original, because it cannot reuse the broken compression structures — Compress PDF will bring it back down without touching the pages themselves.

How PDFs get corrupted, and how to stop it

Almost all real-world PDF corruption traces back to a small number of causes, and most of them are avoidable:

• Interrupted transfers. A download that drops, a network share that disconnects mid-copy, or a USB drive pulled out before the write finished. Always eject removable media properly, and verify file size after large transfers.
• Editing a file in cloud storage while it syncs. Saving over a PDF that Dropbox, OneDrive, or Google Drive is currently uploading is a reliable way to produce a file that is half old and half new. Let the sync finish before opening the file.
• Applications crashing during save. A PDF written by a program that quit unexpectedly typically has content but no valid index — the classic repairable case.
• Emailing PDFs as inline attachments through old servers. Some legacy mail systems still apply text-mode encoding to binary attachments, which corrupts them in transit. Zip the file before sending if you are dealing with a system you do not trust.
• Failing storage. Bad sectors and dying memory cards corrupt files silently. If more than one file on a drive has gone bad, the drive is the problem — copy everything off it before doing anything else.

Frequently asked questions