Guide
How to export an accessible PDF from Word
The export path that keeps tags and the three that silently destroy them, plus what still needs fixing in Acrobat afterwards.
About 9 minutes · quick
The one thing that matters
Word produces genuinely good tagged PDFs. Almost every “Word made an inaccessible PDF” problem is not Word — it is the export path.
Use File → Save a Copy (or Save As) and choose PDF. Then open Options and confirm:
- ☑ Document structure tags for accessibility
- ☑ Create bookmarks using: Headings
- ☐ Bitmap text when fonts may not be embedded — off. This rasterises text.
- ☐ PDF/A compliant — off unless you need archival conformance, because PDF/A-1 forbids some things PDF/UA needs.
Never use File → Print → PDF. Not Microsoft Print to PDF, not Adobe PDF as a printer, not macOS’s Save as PDF from the print dialog. All three go through a print pipeline that discards structure. The output looks identical and has no tags.
On macOS the distinction is easy to get wrong because the Print dialog’s PDF menu is so convenient. It is the wrong button.
Before you export
Five minutes here saves an hour in Acrobat.
Headings must be real heading styles. Bold 16pt text is not a heading; Heading 1 is. Use the Styles gallery, and use the levels in order — no jumping from H1 to H3, which is condition 14-003. If the built-in styles are ugly, modify them rather than abandoning them.
Lists must be real lists. The list buttons, not a hyphen and a tab. A hand-made list exports as a sequence of paragraphs that happen to start with dashes, and a screen reader will not announce “list, 5 items”.
Tables need a header row. Select the top row → Table Design → Header Row, and in Table Properties → Row, tick Repeat as header row at the top of each page. That checkbox is what becomes THead in the PDF, and without it a table spanning three pages has headers on page one only.
Also, in Table Properties → Row, untick Allow row to break across pages. Rows split across a page boundary export as two rows.
Every image needs alt text or an explicit decorative flag. Right-click → View Alt Text. If the image is decorative, tick Mark as decorative rather than leaving the field empty — that becomes an artifact in the PDF instead of an untagged image, which is condition 01-005.
Set the document title. File → Info → Properties → Title. Not the filename — the actual title, the one you would say out loud. This is the single most common PDF accessibility failure in existence, and it is one field.
Set the language. Review → Language → Set Proofing Language, and for any passage in another language, select it and set its language separately. Word exports this as /Lang per run, which is exactly what WCAG 3.1.2 asks for and what nothing else does automatically.
Run Word’s own checker. Review → Check Accessibility. Fix everything it reports. It is a much better tool than Acrobat’s equivalent.
What Word still gets wrong
Even after a clean export, three things reliably need attention.
The document title. Word writes the title into the PDF’s metadata but does not always set the flag that makes a reader display it instead of the filename. In Acrobat: File → Properties → Initial View → Window Options → Show: Document Title. This is condition 06-001 and it takes ten seconds.
Text boxes and floating shapes. Anything with text wrapping other than “In Line with Text” is appended to the end of the page’s structure rather than placed where it appears. A pull-quote in the middle of page 3 is read after the last paragraph of page 3. If you can, convert text boxes to inline content before exporting. If you cannot, fix the order in Acrobat’s Tags panel afterwards.
Complex tables. Word emits Scope on header cells, which is correct and sufficient for a simple grid. It does not emit Headers/ID references, so a table with merged cells or two header rows will need those added by hand — see tagging tables.
Verify, do not assume
Export, then check. The two-keystroke test first: open the PDF, select all, copy, paste into a plain text editor. Then run it through a free checker.
If the paste is empty, you printed instead of exporting. Go back and do it again — there is no way to recover structure from a printed PDF, and re-exporting takes thirty seconds where remediating from scratch takes an hour.
Frequently asked questions
Why did my tags disappear when I made the PDF?
Almost certainly because you printed to PDF rather than exporting. The Microsoft Print to PDF driver, Adobe PDF as a printer, and macOS's Print → Save as PDF all rasterise the layout through a print pipeline that has no concept of document structure. The result is a visually identical file with no tags at all.
Does Word's Accessibility Checker mean the PDF will be accessible?
It means the Word document is in a state where a good export can produce a good PDF. It says nothing about the export itself, and the export is where most of the failures are introduced.
Do I still need Acrobat after a clean export?
Usually for two things: the document title in the Properties dialog, which Word fills with the filename, and the reading order of anything in a text box or a floating shape, which Word appends at the end of the page rather than in place.
Is Google Docs' PDF export tagged?
Yes, since 2022, and reasonably well for headings and lists. Tables come out weaker than Word's, and it has no equivalent of Word's heading-to-bookmark mapping.
Related failure conditions
- 01-005 Content is neither marked as Artifact nor tagged as real content Every text-showing, image-painting and path-painting operator on a page must sit either inside a marked-content sequence that maps to the structure tree, or inside an artifact sequence. Content that is in neither is orphaned.
- 06-001 Document does not contain an XMP metadata stream The catalog has no /Metadata entry pointing at an XMP packet. PDF/UA-1 requires one — it is where the conformance claim and the document title live.
- 14-001 Headings are not tagged Text that is visually a heading — larger, bolder, isolated — is tagged as a paragraph. The document has no navigable outline.
- 11-001 Natural language for text in page content cannot be determined No /Lang is in effect for some text: neither on the catalog, nor on an enclosing structure element, nor on a marked-content sequence.
- 15-003 A TH cell does not contain a Scope attribute Header cells must declare whether they head a row, a column, or both, unless the table instead uses explicit /Headers ID references on every data cell.
Check your own file. Taggart's validator is free and unlimited — no page cap, no watermark, no account.
Last updated .