Guide
PDF accessibility checklist
Use this list to triage a PDF before publishing. The first group can be checked by software. The second needs a person, every time.
Checks software can make
- Title set in document properties and shown in the title bar (WCAG 2.4.2; PDF18).
- Default language declared (WCAG 3.1.1; PDF16).
- The file is tagged (PDF/UA-1 requires a structure tree).
- Pages contain real text, not only images of text (WCAG 1.4.5).
- Images have some alternate text attribute present (presence only).
- No encryption setting that blocks assistive technology.
Checks that need a person
- Is the reading order logical, including sidebars and multi-column layouts? (1.3.2)
- Do headings reflect the real outline, without skipped levels used for styling? (1.3.1)
- Is each alt text accurate and useful, and are decorative images marked as artifacts? (1.1.1)
- Do tables have correct header cells and simple structure? (1.3.1)
- Is colour contrast sufficient and is colour not the only cue? (1.4.3, 1.4.1)
- Are link texts meaningful out of context? (2.4.4)
Worked example
Software would report two failures and two passes. A person still has to judge reading order and the chart description. A tool reporting 'four of six checks passed' says nothing about whether the report is usable.
Title ............ missing -> automated: FAIL (fixable from metadata) Language ......... missing -> automated: FAIL (fixable) Tagged ........... yes -> automated: PASS Text selectable .. 6/6 pages-> automated: PASS Reading order .... 2-column -> manual: NEEDS REVIEW Chart on p.4 ..... alt='chart' -> manual: alt present but not meaningful
Limitations
No checklist covers every success criterion for every document. Automated tools report on what they test. Use the results to prioritise, not to declare a file accessible.
Sources
Check your own PDF
Run a free automated check to see which of these issues appear in your file and whether any are repairable.
Last updated 2026-10-01