To remove PDF metadata properly, the file has to be rewritten, not edited in place. A lot of tools clear the Author and Title boxes you can see in document properties and call it done, while the same values sit untouched in the XMP packet a few hundred bytes further down. If you have ever stripped a PDF and then watched the original author name reappear in a different reader, that is what happened.

The short answer

Open tools.geotaggingimage.com/pdf/metadata, drop the file in, review what it found, and choose to remove everything. The tool rebuilds the PDF and hands back a new file with no document information dictionary entries and no XMP packet. Nothing leaves your browser. If you prefer a command line, exiftool -all= file.pdf does the same job with one caveat covered below.

Then read the last section, because metadata removal does not touch anything that is visibly printed on the page.

Where PDF metadata actually lives

A PDF can carry the same piece of information in three separate places, which is the root of the problem.

Location	What it holds	Typical viewer that shows it
Document information dictionary	Title, Author, Subject, Keywords, Creator, Producer, CreationDate, ModDate	Acrobat’s Document Properties, most readers
XMP metadata packet	An XML block mirroring the above, plus tool history, edit timestamps, document IDs, sometimes original filenames	Acrobat’s Additional Metadata, exiftool, Preview on macOS
Object-level and hidden data	Embedded file attachments, form field values, JavaScript, bookmarks, comments, deleted-but-retained page objects	Nothing, until someone goes looking

The information dictionary is the easy one. Any library can overwrite it. The XMP packet is where the interesting residue usually is: which version of Word or InDesign made the file, when it was last saved, the xmpMM:DocumentID that links this file to every other export from the same source document, and often the account name of whoever ran the export.

Incremental updates make it worse

PDF supports incremental saving. When an editor changes a document, it can append the changes to the end of the file and leave the old bytes in place. The reader follows the newest cross-reference table and shows the current version. The previous version is still there.

So a PDF can hold three generations of author names, none of them visible in a reader and all of them readable in a text editor. Run strings over a suspect file and search for a name you expect to find. If it appears more than once, you have update history.

Hiding versus removing

Here is the distinction that matters when you compare tools.

Hiding sets the visible fields to empty strings. The dictionary now reads /Author (). The XMP packet is untouched. The file length barely changes because nothing was rebuilt. Any tool that returns your PDF in a fraction of a second and keeps the byte count nearly identical is doing this.

Removing parses the document structure, drops the metadata objects, and serialises a fresh file from the remaining page tree. The old cross-reference tables go away because the new file has only one. Byte count changes, sometimes by a lot. This is what our PDF metadata tool does, using pdf-lib in the browser.

A test you can run on any tool: strip a PDF, then open the output in the metadata viewer again, or run exiftool -a -G1 output.pdf. The -a flag shows duplicate tags and -G1 shows which group each one came from. If you see [XMP-xmp] or [XMP-pdf] entries listing values you thought you deleted, the tool hid them.

Removing PDF metadata in the browser

Go to tools.geotaggingimage.com/pdf/metadata.
Drop the PDF onto the drop zone, or click to pick a file. It is read locally with pdf-lib. There is no upload step.
The page lists every field it found, split into document information and XMP. Read this before you clear anything. It is often the only time you will see what a client or an employer left in a file.
Choose Remove all metadata, or clear individual fields if you want to keep, say, the title but drop the author.
Download the rewritten file.

Free accounts get a daily operation cap and 5 files per batch. Pro raises the batch to 50 and drops the cap. No account is needed, and there is no watermark on the output.

If you want to confirm the no-upload claim rather than take our word for it, open DevTools, switch to the Network tab, and process a file. Or turn off your wifi and work offline. The tools keep running because the processing code is already in the page. The same approach runs our photo metadata remover and EXIF viewer.

The Acrobat equivalent

Acrobat Pro has a dedicated feature for this, and it goes further than the properties dialog.

Open the PDF in Acrobat Pro.
Go to Tools, then Redact, then Sanitize Document.
Acrobat warns that all hidden information will be removed and that the change cannot be undone. Click OK.
Save under a new filename.

Sanitize Document removes metadata, embedded search indexes, file attachments, JavaScript, form actions, hidden layers and overlapping objects. Editing the fields in File then Properties then Description does none of that. It writes new values into the information dictionary and leaves everything else alone. Acrobat Reader, the free one, cannot do either: it shows properties read-only.

The exiftool command line

ExifTool handles PDFs, with one important quirk.

exiftool -all= document.pdf

That writes document.pdf and keeps the original as document.pdf_original. The quirk: by default ExifTool edits PDFs using an incremental update. The old metadata is still physically in the file, and ExifTool will tell you so. Adding the -a and -G1 flags on a later read will still surface it in some cases, and exiftool -pdf-update:all= file.pdf is what reverses the operation, which tells you the original data survived.

To get a real removal from the command line, rewrite the file afterwards:

exiftool -all= document.pdf
qpdf --linearize document.pdf clean.pdf

The second step is the one that actually drops the old bytes.

What removing metadata does not do

This is the part most guides leave out, and it is the part that causes real damage.

It does not remove text on the page. If the document has a footer reading “Prepared by J. Wilson, Draft 3”, that is page content, not metadata. Stripping metadata leaves it exactly where it is.

It does not redact anything covered by a black box. Drawing a filled rectangle over a name in a PDF editor adds a shape on top of the text. The text object underneath is unchanged. Select the area, copy, and paste into a text editor and the name comes back. This has burned law firms, government agencies and newsrooms repeatedly. Real redaction deletes the underlying content stream, which is what Acrobat’s Redact tool does before you sanitize, and what our tools do not do.

It does not clean embedded images. A PDF built from photos can carry the original EXIF inside each embedded JPEG. If you assembled the file from camera images, strip the photos first with the metadata remover, then build the PDF. Our photo metadata guide covers what those images carry.

A cleaned file still has a processing date. Any tool that writes a new PDF creates a new file with a modification timestamp, and most writers set a fresh ModDate. You cannot produce a PDF with no dates at all and still have a valid file. What you can do is make sure the only date present is the one from the moment you cleaned it, which reveals nothing about the document’s history.

A checklist before you send

Run the file through the PDF metadata tool and read the report before removing anything.
Search the visible text for names, draft markers, internal codes and email addresses.
Select-all and copy the page into a plain text editor to catch anything sitting under a black box.
If the PDF came from merged sources, remember that merging can carry metadata from the first document forward, so remove metadata after merging, not before.
Check embedded attachments. Acrobat lists them in the paperclip panel.

Metadata removal is quick. The page content review is the slow part, and it is the one that decides whether the file is actually safe to send.

Remove PDF Metadata: How to Actually Delete It