PDF troubleshooting has a bad reputation because most advice about it is a list of things to try. Try a different reader. Try re-downloading. Try converting it to something else and back.
That works eventually and teaches you nothing. Almost every PDF problem has one of four underlying causes, and a minute spent identifying which one you have replaces an afternoon of trying things.
Key facts
- Will not open at all
- Usually a damaged cross-reference table - repair it
- Opens but text will not copy
- Usually no text layer - it is a scan
- Pages appear blank
- Damaged content streams, or genuinely blank pages
- Fonts look wrong elsewhere
- Fonts were not embedded
- Enormous file
- Embedded images, almost always
- Prints differently from screen
- Scaling, or an unembedded font
The one-minute diagnosis
Run these in order. The first one that fails tells you what you are dealing with.
- Does it open?
If no reader will open it, the file structure is damaged - typically an interrupted download. Repair PDF rebuilds the index. If the repair reports fewer pages than you expected, the rest never arrived.
- Can you select a word?
If not, the document has no text layer. It is a scan, and search, extraction and redaction will all return nothing. This is not a fault; it is what the file is.
- Do the pages show content?
Blank pages in an otherwise working file mean damaged content streams, which a structural rebuild sometimes recovers. They can also mean the pages really are blank, which a scanner produces for every unprinted reverse side.
- Does it look right to you but wrong to someone else?
An unembedded font. The other machine is substituting, which shifts spacing and line breaks and can turn symbol fonts into boxes.
- Is it unreasonably large?
Embedded images. A scanned page is several megabytes on its own. Compress it, and read the reported saving - a small saving means the weight was never images.
Tools for each diagnosis
Files that will not open
A PDF keeps a cross-reference table: an index of where every object in the file lives. Readers use it to find the pages. When it is damaged, a reader that cannot locate the pages refuses the file entirely, even though most of the content is sitting there intact.
The overwhelmingly common cause is a download that stopped part-way. The file is nearly the right size, it looks complete in a folder listing, and nothing will open it. A structural repair rebuilds the index from whatever objects survived, and recovers such files often enough to be the first thing to try - after simply downloading it again, which is better than any repair.
Text that will not copy
If selecting text produces nothing, there is no text to select. The document is an image of a page. Every downstream complaint - search finds nothing, extraction returns an empty file, redaction refuses - follows from that single fact.
The distinction is invisible on screen, which is why this catches people repeatedly. A scan and a generated PDF look identical. Only selection tells them apart.
What to do with a scan
You can read it, print it, crop it, rotate it, split it and compress it. You cannot search it, extract text from it, or redact it. Converting it to an image format changes nothing, because it already is one. Only optical character recognition would make it text, and this site does not offer that rather than offering it badly.
Symptom to cause
The five that account for most of it.
| Symptom | Most likely cause | Fix |
|---|---|---|
| No reader will open it | Damaged cross-reference table | Re-download; failing that, repair |
| Text cannot be selected or searched | No text layer - it is a scan | Nothing here; OCR is a different job |
| Pages render blank | Damaged content streams, or genuinely blank | Repair, then check the page count |
| Layout differs on another machine | Font not embedded | Re-export from the source with fonts embedded |
| File is 80 MB for 20 pages | Embedded scan images | Compress, or rescan in greyscale |
Documents that print differently from how they look
Two causes, and they are easy to tell apart. If the content is the right size but positioned differently, a font was substituted and the line breaks moved. If everything is proportionally smaller, the print dialogue scaled it - 'fit to page' is the default on many systems and it shrinks the document by a few per cent.
The second is far more common and matters far more, because a few per cent of scaling makes every measured dimension on a graph wrong while looking completely normal.
Before concluding a file is broken
Open it in a second reader. A surprising proportion of 'broken PDF' reports are a single reader disagreeing with an unusual but valid file - most often one using a feature the reader does not implement.
If two independent readers both refuse it, the file is genuinely damaged. If one opens it, use that one and move on.
When a document is not behaving
In this order.
- Try a second reader before anything else
- Re-download from the original source if you still can
- Try to select a word, to establish whether there is a text layer
- Check the page count against what you expected
- Run a structural repair if it will not open
- Check the file size against what the content should weigh
- Look at the document properties for the producing software, which often explains the oddity
Guides in this topic
Every one written by a named author and checked against real documents.
When a PDF looks different elsewhere
The format exists to prevent exactly this. When it happens anyway there are three causes, and you can identify which in about a minute.
When PDF pages are blank
Four causes, and they need four different responses. The one that matters most is the one where the pages are blank on paper but not on screen.
Why a PDF is enormous
Almost always images. Working out which images, and whether anything can be done about them, takes two minutes and prevents a lot of pointless compressing.
Why text will not copy
Three causes, and the first one accounts for most of them. All three are diagnosable in under a minute and only two have anything you can do about them.
When a PDF will not open
Four causes, in order of likelihood, with a test for each. Most of them are recoverable and one of them is not, and you can tell which in about a minute.
Questions and answers
My PDF says it is damaged. Can it be fixed?
Often, if the damage is structural. A rebuild reconstructs the index from the objects still in the file. It cannot recover content that never finished downloading, and the page count in the result tells you which situation you are in.
Why can I see the text but not select it?
Because what you can see is a picture of text. The document is a scan, and it contains no characters at all. This is the single most common PDF misunderstanding.
Why is my 20-page PDF 80 MB?
Embedded images, almost certainly from scanning. Compress it and check the reported saving; a scanned document typically drops by seventy per cent or more.
A document looks wrong on my colleague's machine. Why?
A font that was not embedded in the file, so their reader substitutes the nearest match and the layout shifts. Re-export from the source application with font embedding turned on.
Some pages are blank when I print but not on screen.
Usually annotations or form fields that are displayed but not set to print. Flatten the document, which draws them permanently into the page content, and print again.