PDF & Word Metadata Remover

Every PDF and Word file carries a little dossier about itself: who wrote it (frequently a real account name), what company license produced it, which software and version, when it was created and every time it was touched since. Most people have sent a "clean" document that quietly named a colleague, a client, or a template path. The first thing this page does is show you that dossier - for your file, exactly as written. The second thing it does is remove it, in your browser: PDFs are blanked in place so the file stays byte-for-byte structurally valid, Word files get their properties parts swapped for empty ones with the document itself copied untouched.

One honest boundary, stated plainly on the page and in the FAQ: metadata removal is not redaction. This cleans the properties about the document; it does not touch text inside the document. And where a PDF stores its info dictionary somewhere in-place blanking can't reach, the tool says what it could not clear instead of pretending. Files that would need guessing are refused, not mangled.

How to use the metadata remover

1. Click Choose a .pdf or .docx file, or drag the file onto the drop area.
2. Read the table. Each row lists a field and the value that was found, crossed out because it has been removed from the copy. Warnings appear below the table.
3. Click Download cleaned copy. The file is saved with -clean added to its name. Your original is not changed.
4. To check the result, drop the cleaned copy back in. A cleaned PDF still lists the same fields but with empty values, because the fields are still there with blank contents. A cleaned Word file should report that no metadata values were found.

Where metadata lives in a PDF

A PDF can keep document metadata in two places.

• The Info dictionary, which the file trailer points to. Its standard keys are Title, Author, Subject, Keywords, Creator (the app that made the original), Producer (the app that wrote the PDF), CreationDate, and ModDate. Values are literal strings in parentheses or hex strings in angle brackets. Dates look like D:20260914101500-04'00'. Programs can add their own keys.
• The XMP stream, an XML packet the document catalog points to. It is wrapped in <?xpacket begin and <?xpacket end markers and repeats much of the same data, such as dc:creator and xmp:CreatorTool.

Since PDF 1.5, objects can also sit inside compressed object streams. And an incrementally saved PDF often keeps older versions of these objects earlier in the file.

This tool edits the bytes in place so every offset in the cross-reference table stays valid. It searches the whole file, including older copies, for the eight standard keys above followed by a string value. Literal values are replaced with spaces, and hex values with zeros. It also replaces everything between xpacket markers with spaces. It does not touch custom Info keys, the trailer's /ID, XMP inside a compressed stream, annotation authors, or form field values. Encrypted PDFs are refused.

For .docx files, the parts docProps/core.xml, docProps/app.xml, and docProps/custom.xml are replaced with empty versions. Everything else in the zip is copied byte for byte.

Worked example

An Info dictionary written by a Word export:

9 0 obj
<< /Title (Q3 Pricing (draft)) /Author (jdoe) /Company (Acme Corp)
   /Creator (Microsoft Word) /Producer (Microsoft Word for Microsoft 365)
   /CreationDate (D:20260914101500-04'00') >>
endobj

The table lists Title, Author, Creator, Producer, and CreationDate, and the status reports 5 values blanked. In the cleaned copy the same object reads:

9 0 obj
<< /Title (                  ) /Author (    ) /Company (Acme Corp)
   /Creator (              ) /Producer (                                )
   /CreationDate (                       ) >>
endobj

/Company is a custom key, so it stays.

Common problems and fixes

• Bookmark names are blank in the cleaned PDF. Cause: bookmark entries also use a /Title key, and the byte scan cannot tell them apart from the document title. Fix: if you need the bookmarks, keep the original and export a fresh PDF from the source app with its document properties cleared.
• Another viewer still shows the author or company. Cause: a custom Info key, XMP in a compressed stream, or an Info dictionary inside an object stream. The tool only warns about that last case when it found nothing else. Fix: clear the properties in the source app, export again, and run the new PDF through here.
• "This PDF is encrypted" for a file that opens without a password. Cause: a PDF with only a permissions password still contains an encryption dictionary. Fix: remove the restrictions in a PDF editor you are authorized to use, then clean the result.
• Names still show up in a cleaned Word file. Cause: comment authors and tracked-change authors are stored in the document body, not in the properties parts. Fix: in Word, use File, Info, Check for Issues, Inspect Document, and remove comments, revisions, and personal information.
• A .doc file is rejected. Cause: only .docx is supported. The older .doc format is a different binary container. Fix: save it as .docx in Word first.

Related tools

If the PDF is a statement you need as a spreadsheet, the PDF to CSV converter extracts its text rows.

Questions this tool gets asked

What metadata hides in a PDF or Word document?

More than people expect: the author's name (often a real login name), company, title, the software and version used, creation and modification dates, and sometimes template paths. This page shows you exactly what your file carries before anything is removed.

How does the removal work - is my document uploaded?

Never. The file is edited by JavaScript in your tab. For PDFs, the metadata values are blanked in place so the file structure stays byte-for-byte valid; for Word files, the document properties parts are replaced with empty ones. The cleaned file downloads straight from browser memory.

Does this also redact the document's contents?

No - and it is important to be clear about that. This tool removes the properties ABOUT the document (author, dates, software). Text inside the document, including anything under a black rectangle, is untouched. Never rely on drawn boxes for redaction.

What exactly is removed from a Word file?

The core properties (author, last modified by, title, dates, revision number), the app properties (software, company, editing time), and every value in the custom properties part. The document text, images, and formatting are copied over untouched.

Are there PDFs it can't fully clean?

Yes, and it tells you when. Encrypted PDFs are refused outright. Some PDFs store their info dictionary inside compressed object streams where in-place blanking can't reach - the tool reports what it found and what it could not clear, rather than claiming a clean it didn't do.

How do I just see what metadata a document carries, without changing it?

Drop it here and read the table - the values are shown before anything downloads, and your original file on disk is never modified (the clean copy is a separate download). Adobe Reader's File > Properties and Word's File > Info show some of the same fields, but not the XMP packet or custom properties this page surfaces.

Related tools

• pdf → csvPDF tables out as CSV - bank statements never leave your machine.
• msg / eml → readable emailOpen Outlook emails - body, headers, attachments - without Outlook.
• sha hashSHA-256, SHA-1, and SHA-512 digests of any text, computed locally.
• statement → csvStatement PDF in, clean transaction CSV/Excel out - never uploaded.

Printable templates

• medical consentLets a caregiver authorize treatment when the parent cannot be reached.
• massage intake formClient intake with the health-history checklist, printed under your letterhead.
• liability waiverRelease form with the acknowledgments that make a waiver mean something.