Get a fresh copy first if the file was damaged on its way to you; a clean download fixes that completely. Otherwise, most corruption is a damaged cross-reference table, the index at the end of the file, and the pages are usually fine and simply cannot be located.
Repair works by scanning the file for valid objects and rebuilding the index. When the damage is limited to the table, recovery is often complete.
The symptoms vary: a file that will not open at all, one that opens to blank pages, one that shows the first few pages and then stops, or an error about a damaged or malformed document. All of them usually come back to the same underlying issue. If you are not sure the file is damaged at all, the four reasons a PDF will not open covers the causes that are not corruption.
Why PDFs break this way
A PDF is a collection of numbered objects, and the reader needs to find any of them immediately without scanning the whole file. That is what the cross-reference table provides: a list of every object and its exact byte position, stored at the end.
Keeping it at the end is what makes incremental saving possible, since changes can be appended with a new table. It also means the most critical structure in the file sits in the most vulnerable place. A transfer that stops early, a disk error in the final bytes, or a tool that writes the table incorrectly leaves a file whose content is entirely intact and completely unreachable.
The cross-reference table at the end, not the page content.
Scans the file for valid objects and constructs a new table from what it finds.
When only the table was damaged, the recovered document is identical to the original.
Where the damage comes from
- Interrupted downloadFile is truncated
- Interrupted transfer or copySame, at the end
- Email gateway modificationBytes altered in transit
- Faulty storage mediaRandom damage anywhere
- Buggy generatorWritten wrong from the start
- Editing crash mid-savePartial write
What repair does
A repair tool ignores the broken index entirely and reads the file from the beginning, looking for anything that has the shape of a valid PDF object. Each one it finds gets catalogued. At the end it assembles the surviving objects into a fresh document with a correct table.
How complete the result is depends on how much survived. A file that lost only its table recovers fully, because every object is still present and merely unlisted. A file with damage running through the page objects recovers the undamaged ones and loses the rest, which usually shows up as missing or partial pages.
Structure damaged, content intact
- A corrupted or missing cross-reference table
- Truncation that removed only the index
- Incorrect object offsets from a faulty writer
- A damaged trailer at the end of the file
Content itself is damaged
- Truncation that removed actual page data
- Random corruption through the middle
- Damaged compressed streams
- Encrypted files with damaged key data
- Files that were never valid PDFs
What to try, in order
Ask for the file again
Unglamorous, fast, and reliable whenever the damage happened on the way to you. Request a fresh copy, ideally through a different route, since email gateways and chat applications sometimes modify attachments in transit.
Confirm it is a PDF
Open it in a text editor and look at the first few characters. A valid PDF begins with a percent sign followed by PDF and a version number. Anything else means the file is something other than what its name claims.
Try a different reader
Readers vary considerably in how much damage they tolerate, and several will silently rebuild an index that another refuses to work with. A browser is worth trying, since browser engines are quite forgiving.
Run a repair tool
Once the simpler options are exhausted, run the file through Repair PDF. It rebuilds the index from the objects that survived, and if the structure is beyond saving it redraws every page that can still be drawn and builds a new file from those.
The exact message, and what it tells you
Readers are not consistent about wording, but a handful of messages come up often enough to be worth reading literally. Each is produced at a different point in the process of opening a file, and knowing which point tells you how much of the document is likely to survive.
"The file is damaged and could not be repaired"
Acrobat shows this after it has already run its own recovery pass and got nowhere. That second clause matters: the reader is not declining to try, it is reporting that trying did not work. The message covers three quite different situations.
A common one is that the file is not a PDF. A download that failed part way through often leaves a saved error page carrying the .pdf name, and a reader opening it finds no PDF header at all. This takes ten seconds to check. Open the file in a plain text editor and look at the first line: a real PDF begins with %PDF-1. and a digit, while an error page begins with something like <!DOCTYPE html>. If that is what you have, no repair tool will help and downloading it again will.
The second is honest truncation, where the transfer stopped part way and everything that did arrive is real PDF. A recovery pass can usually rebuild whatever made it, and how much that is depends on where the transfer stopped. The third is damage inside the object data itself, and it is the case that cannot be salvaged.
"This file cannot be opened because it has no pages"
This message is more specific and, oddly, more hopeful. The reader got far enough to parse the document catalogue, so the structure is broadly intact. What it could not find was the page tree, the object that lists which pages exist and in what order.
Files arrive in this state when an export was interrupted after the document was created but before its pages were attached, or when a generator wrote a catalogue pointing at a page list that never got written. It also appears when a portal hands back an error response that happens to be a technically valid PDF with nothing inside it.
Because the page objects themselves are often still sitting in the file, a recovery pass that scans for them rather than following the index can frequently rebuild the tree and produce something that opens. If the pages were never written, no tool can invent them, and what you have is a few kilobytes of wrapper.
Encrypted files
Damage to an encrypted PDF is harder, because the recovery process needs to make sense of the structure before it can decrypt anything, and the encryption information is itself part of the structure that may have been damaged.
If the encryption dictionary survived and you have the password, repair can often proceed. If that part of the file was damaged, the content is effectively unrecoverable even though the encrypted bytes are still present, because nothing knows how to decrypt them.
Repair a damaged PDF
Structure rebuilt where that is possible, pages re-rendered where it is not. Nothing uploaded.
Common questions
What can and cannot be recovered.
01What does it mean when a PDF is corrupted?
Usually that the cross-reference table, the index of object positions stored at the end of the file, is damaged or missing. The pages themselves are often completely intact and simply cannot be located by the reader.
02How does repair work?
The tool ignores the broken index and scans the file from the start for anything shaped like a valid PDF object. It catalogues what it finds and builds a fresh document with a correct index from the surviving objects.
03Why do PDFs get damaged so often at the end?
Because the most critical structure in the file lives there. Keeping the index at the end is what allows incremental saving, and it also means an interrupted transfer removes the one part the reader cannot manage without.
04Can every corrupted PDF be repaired?
No. Files that lost only their index usually recover completely. Files where the damage runs through the page content itself recover partially at best, since repair can only rebuild an index to objects that still exist.
05What should I try first?
Check the file size against the source, and ask for a fresh copy. Truncated downloads are a very common cause and re-downloading fixes them entirely, in far less time than any repair attempt takes.
06Can I repair an encrypted PDF?
Sometimes. If the encryption information survived and you have the password, repair can often proceed. If that part of the structure was damaged, the content cannot be decrypted even though the bytes are still in the file.
07What does "the file is damaged and could not be repaired" mean?
That Acrobat tried its own recovery pass and failed. Before trying another reader, drop the file into Repair PDF: its first check reads the header and says whether the file is a saved web page rather than a PDF, in which case only a fresh download will help.
08What does "this file cannot be opened because it has no pages" mean?
The document catalogue parsed but the page tree is missing or empty. The page objects are often still in the file, so a recovery pass that scans for them rather than following the index can frequently rebuild it.