How to Make a PDF Fully Accessible for Screen Readers (PDF/UA & Section 508 Guide)

A screen reader can only read a PDF properly if the file is tagged. Tags are a hidden structure layer that says which text is a heading, a paragraph, a list item or a table cell, in what order it should be read, and what each meaningful image shows. Making a PDF accessible means creating that structure, ideally by exporting correctly from the source document, then testing the result with a checker and with a real screen reader.

PDF/UA (ISO 14289) is the standard that sets out the technical requirements for an accessible PDF. WCAG, the guideline that Section 508 and similar rules refer to, adds broader criteria such as color contrast and meaningful link text. A file can pass every PDF/UA check and still fail WCAG, so plan for both.

What a screen reader does with an untagged PDF

Without tags, NVDA, JAWS or VoiceOver read text in the order it happens to be stored in the file. That order often has nothing to do with the layout. Two columns get read across instead of down, the running header is announced on every page, a table becomes a stream of unconnected numbers, and images are skipped or announced only as "graphic". Some readers try to guess the structure on the fly, with mixed results.

A scanned PDF is worse: each page is a photograph of text, so there is nothing to read at all. That needs OCR before any tagging can start, covered further down.

What an accessible PDF needs

• ●Tags on all meaningful content, with decorative items such as repeated page numbers, rules and background shapes marked as artifacts so they are skipped.
• ●A logical reading order that matches how a sighted person reads the page.
• ●Real headings tagged H1 to H6, in order, without skipping levels.
• ●Alt text on informative images, and decorative images marked as such.
• ●Tables with header cells tagged as headers, so a reader can announce "Q3, Revenue" instead of a bare number.
• ●Lists tagged as lists, so the reader announces how many items there are.
• ●A document title in the metadata, set to show in the window title bar instead of the file name.
• ●The document language set, for example English (United States), so the speech engine pronounces words correctly.
• ●Fonts embedded with proper character mapping, so the reader gets real letters rather than symbols.
• ●Form fields with descriptions (tooltips) that say what to enter.
• ●Bookmarks for long documents, generated from the headings.

Start in the source document, not the PDF

Repairing tags in a finished PDF is slow and fiddly. Generating them from a well-structured Word or LibreOffice file takes seconds, so most of the work happens before you export.

Microsoft Word for Windows

1. 1.Format headings with the built-in Heading 1, Heading 2 and Heading 3 styles. Bold, larger text is not a heading to a screen reader.
2. 2.Make lists with the Bullets and Numbering buttons, not typed dashes or numbers.
3. 3.Right-click each image and open the alt text option. Describe what the image tells the reader, or tick Mark as decorative.
4. 4.In tables, select the first row, open Table Properties, and on the Row tab tick Repeat as header row at the top of each page. This also marks the row as a header.
5. 5.Set the title in File > Info > Properties, and check the proofing language under Review > Language.
6. 6.Run Review > Check Accessibility and fix what it lists.
7. 7.Choose File > Save As (or Save a Copy), pick PDF as the file type, click Options, and make sure Document structure tags for accessibility is ticked. Set Create bookmarks using to Headings while you are there.

Word for Mac

Prepare the document the same way, then choose File > Save As, set File Format to PDF and select Best for electronic distribution and accessibility (uses Microsoft online service). That option produces the tagged version. The other option, Best for printing, does not.

LibreOffice Writer

Use paragraph styles for headings and add alt text through each image's properties. Run Tools > Check Accessibility, then choose File > Export As > Export as PDF and, on the General tab, tick Universal Accessibility (PDF/UA). Tagging is switched on automatically when that option is selected.

Repairing a finished PDF in Acrobat Pro

When you only have the PDF, you need an editor that can work with tags. Acrobat Reader cannot add them. In Acrobat Pro, the tools sit under All tools > Prepare for accessibility (in older versions, Tools > Accessibility).

1. 1.Automatically tag PDF if the file has no tags. If it already has some, Acrobat asks before replacing them.
2. 2.Check for accessibility runs the built-in checker and lists problems by category in a panel on the right.
3. 3.Fix reading order lets you draw and retype regions as headings, text, figures or tables, and mark headers and footers as background so they are skipped. The Order panel lets you drag content into the right sequence.
4. 4.Set alternative text steps through every figure so you can type a description or mark it decorative.
5. 5.For tables, use the table editor in the reading order tool to mark header cells.
6. 6.Open File > Properties. On the Description tab, enter a title. On the Initial View tab, set Show to Document Title. On the Advanced tab, set the Language.
7. 7.Save, then run the checker again.

Scanned PDFs: get text into the file first

A scan has no text for tags to point at, so run OCR first. OCR PDF adds an invisible, searchable text layer in your browser using Tesseract, without uploading the document. It does not tag the result, so open the OCR'd file in Acrobat Pro and work through the repair steps above. Proofread the recognized text too: a screen reader reads OCR mistakes exactly as they appear, so "Arnount due" is what the listener hears.

Testing: a checker and your own ears

• ●PAC (PDF Accessibility Checker). A free Windows tool, now distributed by axes4, that runs the machine-testable PDF/UA checks and the WCAG checks that can be automated. It also shows the tag structure so you can see what a screen reader will get.
• ●The Matterhorn Protocol. Published by the PDF Association, it lists every PDF/UA failure condition and marks which ones need human judgment, such as whether alt text is accurate or the reading order is logical.
• ●A real screen reader. NVDA is free on Windows and works with Acrobat Reader. Press H to jump between headings and T to jump to tables. On a Mac, turn on VoiceOver with Command+F5. Listen to the first few pages and to every table.

Things that quietly break accessibility later

• ●Editing pages with the wrong tool. Many merge, split and page-editing tools copy only the visible page content and leave the tags behind. Check the result after any such edit, or make the change in the source document and export again.
• ●Flattening or converting pages to images. Both remove the text entirely, along with any chance of reading it.
• ●Restrictive security settings. If you add a permissions password, make sure text access for screen readers stays allowed. PDF/UA requires it.
• ●Running another export over an accessible file. Re-saving through a print driver or an online converter often produces an untagged copy that looks identical.