Text and document setup
- Text is selectable, searchable, and accurate after any OCR process.
- The document has a useful title and the title is shown in the viewer.
- The default language is set, with language changes marked where needed.
- Security settings do not block access by assistive technology.
Tags and reading sequence
- The PDF has a tag tree with one logical document root.
- Headings, paragraphs, lists, tables, links, and figures use suitable tags.
- Heading levels describe the document hierarchy without unexplained jumps.
- The tag order follows the intended reading order across columns and callouts.
- Decorative and repeated page furniture is marked as artifact content.
- Every piece of meaningful page content belongs in the structure tree.
Images, charts, and other visual information
- Informative images have alternatives written for their purpose here.
- Complex charts have a nearby explanation or equivalent data presentation.
- Decorative images do not add noise to the reading sequence.
- Meaning does not depend on color, shape, or visual position alone.
- Text and meaningful graphics have sufficient contrast.
Tables, links, and forms
- Tables are used for data and have a simple, regular structure where possible.
- Header cells are marked and associated with the correct data cells.
- Links have meaningful text, a link annotation, and a link structure element.
- Form controls expose a name, role, value, state, and clear instructions.
- Required fields and errors are communicated without relying only on color.
- Keyboard focus follows a logical order and remains visible.
Manual use and final release
- Headings, landmarks, links, and form fields are usable with a screen reader.
- Reading order makes sense when content is read continuously.
- The document remains usable under zoom and alternative display settings.
- Alternative text communicates meaning instead of repeating nearby text.
- The exact release file has passed the agreed technical and manual checks.
- The report records tool versions, coverage limits, repairs, and retest status.
How to use the checklist
Assign each group to the person who can act on it. Authors should resolve heading, table, link, image, and language problems in the source. A PDF remediator can repair output structure when the source cannot produce it. An accessibility reviewer should complete the checks that depend on meaning and real use.
Record "not checked" instead of pass when evidence is missing. A checklist is useful because it makes work visible, not because every row eventually turns green without explanation.