How to Make a PDF Accessible: Tags, Reading Order, Testing

What makes a PDF accessible?

An accessible PDF is one a person can read without seeing the page: with a screen reader, a braille display, a keyboard instead of a mouse, or text enlarged until the layout has to reflow. The catch is that a PDF, by default, only records where ink goes on a page. For software to read it aloud in a sensible order, the file needs a second, invisible layer called tags that says what each piece of content is.

That is why two PDFs can look identical and behave completely differently. One is announced as “heading level 1, Annual Report,” then paragraphs, then a table with named columns. The other is read as one unbroken stream of words, or as nothing at all because every page is a photograph.

A PDF that works with assistive technology has six things:

• Real text. Characters you can select and search, not a picture of text.
• Tags. Headings, paragraphs, lists, tables with header cells and figures are each labeled as what they are.
• A logical reading order. The tags follow the order a person would read the page, including across columns and sidebars.
• Text alternatives. Every meaningful image has a short description, and decorative lines and page furniture are marked so they get skipped.
• A document language and title. The language tells a screen reader which voice and pronunciation rules to use.
• Usable links and form fields. Link text makes sense out of context, and each field has a label and a sensible tab order.

Which standard applies to you?

Four names turn up in policies and tender documents. They overlap far more than they conflict.

Scroll the table to compare all columns.

For US federal content, the Revised Section 508 Standards require electronic documents to meet the WCAG 2.0 Level A and AA criteria, with four listed exceptions for documents that aren’t web pages. The U.S. Access Board’s rule is the primary source.

If nobody has told you which standard applies, ask whoever will publish the file. With no written policy, aim for WCAG Level AA, the level most policies name.

Start in the source document, not the PDF

Tags are written when you export, from structure that already exists in the source file. If the Word document uses real headings, the PDF gets heading tags. If it uses bold 16-point text that only looks like a heading, the PDF gets a paragraph. Fixing that in the source takes minutes. Rebuilding it inside a finished PDF means retagging by hand in a specialist editor.

1. Use real heading styles. Apply Heading 1, Heading 2 and so on, in order, without skipping levels.
2. Use the list and table tools. Typed hyphens and tab-aligned columns look right and carry no structure. In tables, mark the header row and avoid merged cells where you can.
3. Add alt text to images. Say what the image tells the reader, in a sentence. Mark purely decorative images as decorative.
4. Write link text that stands alone. “Download the 2026 fee schedule” works. “Click here” doesn’t.
5. Set the document title and language in the file’s properties.
6. Run the built-in checker. In Word and PowerPoint it is Review > Check Accessibility.

Export with tags switched on

• Word or PowerPoint on Windows: File > Save As > PDF, open Options, and keep “Document structure tags for accessibility” checked.
• Word on a Mac: File > Save As > PDF, then choose “Best for electronic distribution and accessibility.”
• LibreOffice: File > Export as PDF, then check “Tagged PDF (add document structure).”

How to test a PDF with free tools

1. Try to select a sentence. If you can’t, the page is an image and needs text recognition first (see the next section).
2. Check that the file is tagged. In Adobe Acrobat Reader, open File > Properties. The Description tab shows “Tagged PDF: Yes” or “No.”
3. Look at the reading order. In Acrobat Reader, choose View > Zoom > Reflow. The page collapses into a single column in the order the file stores it. If a sidebar lands in the middle of a sentence here, a screen reader will read it there too.
4. Run an automated checker. PAC is a free checker for Windows, and Acrobat Pro has one built in. They list missing alt text, untagged content, tables without headers, and a missing title or language.
5. Listen to it. Open the file with a screen reader (NVDA is free on Windows, and VoiceOver is built into every Mac) and jump from heading to heading. Then put the mouse away and press Tab through every link and form field.

Don’t skip the last step. A checker can confirm that an image has alt text, but not that the text says something more useful than “image1.png.” It can confirm that headings exist, but not that they describe what follows. For multi-column layouts, the W3C’s technique for correct reading order shows what to look for.

Scanned PDFs: add text first, then structure

A scan is a stack of photographs. There are no characters for a screen reader to speak, so the first job is optical character recognition (OCR), which works out the words in each image and stores them as an invisible text layer behind the page.

1. Open OCR PDF and choose your scan: one PDF, up to 50 MB and 100 pages.
2. Set Document language to the language printed on the pages, then select Recognize text. Recognition runs in your browser.
3. Read the recognized text against the page. Names, numbers and anything laid out in columns are where mistakes hide.
4. Choose Searchable PDF to get your original pages back with the text layer added. This step uploads the PDF to our server to build the new file, so it takes files of up to 30 MB.

The result can be searched, copied and read aloud. It is still not tagged: it has no headings, no table structure and no alt text, so it won’t pass a formal check. For a document you have to publish as accessible, go back to the original source file if one exists, or add tags in a desktop editor that supports them, such as Acrobat Pro. The W3C describes the same two stages in its technique for scanned documents, and our guide to how OCR works explains how to get cleaner recognition.

What PDF Logic tools do to a tagged PDF

PDF Logic has no tool for adding tags to a finished PDF or repairing them. It is useful before and after that work, as long as you know which tools leave an existing tag structure alone.

Scroll the table to compare all columns.

The practical rule: finish merging, reordering and splitting first, in the source application where you can, and make the tagged export your last step.

Frequently asked questions

Is a searchable PDF the same as an accessible PDF?

No. Searchable means the file contains real text. Accessible means that text also has structure: headings, reading order, table headers, alt text and a language. Every accessible PDF is searchable, but a searchable scan with no tags is only the first step.

Does PDF/A make a document accessible?

Not on its own. PDF/A is about long-term preservation: embedded fonts, no outside dependencies, predictable rendering. PDF to PDF/A produces PDF/A-2b, the level that preserves appearance and doesn’t require tags. Our PDF vs PDF/A guide explains the levels.

Will a password stop a screen reader from reading my PDF?

It can. A protected PDF carries a set of permissions, and one of them controls whether assistive technology may read the text. Protect PDF always leaves that permission on, even when you switch off copying, so a reader who has the password can still use a screen reader. See how PDF passwords and permissions work.

Can I fix an untagged PDF without the original file?

Yes, but it is manual work. A desktop editor with tagging tools, such as Acrobat Pro, can add tags automatically and then lets you correct them. Heading levels, reading order, table headers and alt text all need a human check. For anything longer than a few pages, it is worth asking the author for the source file first.