Checklist

PDF accessibility checklist: 14 checks for WCAG 2.1 AA

Run these checks on every PDF you publish. The first nine can be automated across a whole folder; the rest need a human.

Updated 8 min read

Use this list with a full checker such as PAC (free, Windows) or Adobe Acrobat Pro, and a screen reader test for important documents. The Section508.gov PDF testing module walks through the manual steps in Acrobat.

Tip: Checks marked Automated are run by PDF Triage for every file in a folder, so you can find the expensive problems before opening a single document.

The 14 checks

  1. Text is real, not a scanned imageAutomated

    Select text or search inside the file. If nothing is selectable, run OCR before anything else.

  2. The document is taggedAutomated

    A tag tree (structure tree) must exist and the document must be marked as tagged.

  3. It has a meaningful titleAutomated

    Set a descriptive title in document properties — not a file name like “Microsoft Word - Doc1.docx”.

  4. The window shows the titleAutomated

    Set “Show: Document Title” (DisplayDocTitle) so screen readers announce the title, not the file name.

  5. The primary language is setAutomated

    For example en-US or es-US. Screen readers use it to pick the right voice and pronunciation.

  6. Headings are taggedAutomated

    Use H1–H6 tags that match the visual hierarchy so people can navigate by heading.

  7. Images have alternative textAutomated

    Every meaningful Figure needs alt text; decorative images should be artifacts.

  8. Data tables have header cellsAutomated

    Use TH for header cells (with scope where needed) so cells are announced with their headers.

  9. Form fields have descriptions and tab order follows structureAutomated

    Each field needs a tooltip/label; pages with fields or links should use structure tab order.

  10. Long documents have bookmarksAutomated

    Documents over ~20 pages should have bookmarks that mirror the headings.

  11. Reading order is logicalManual

    Check with the tag tree or a screen reader: columns, sidebars and footnotes must read in the right order.

  12. Alt text is accurate and usefulManual

    Software can confirm alt text exists, not that it describes the image well.

  13. Color contrast is sufficientManual

    Text needs 4.5:1 contrast (3:1 for large text); information is not conveyed by color alone.

  14. Links are descriptiveManual

    Link text makes sense out of context (“2027 budget (PDF)”, not “click here”).

Why the order matters

Scanned and untagged files are the most expensive to fix — they need OCR and full tagging, usually by a person. A file that only lacks a title and language takes seconds. Triage first so the budget goes where it’s needed: see how to estimate remediation cost.

Check your own PDFsFree bulk triage in your browser — no upload, no sign-up.
Open PDF Triage

Frequently asked questions

Can a PDF pass an automated check and still be inaccessible?

Yes. Automated tools confirm structure exists; they cannot judge whether alt text is meaningful or reading order makes sense. The Matterhorn Protocol marks 47 of its 136 PDF/UA failure conditions as usually requiring human judgment.

What is the fastest win?

Title, language and display-title settings. They take seconds per file and PDF Triage Pro can apply them to a whole batch.