PDF accessibility depends on meaningful structure, readable content and support for assistive technology. This guide explains common requirements and checks; a visual preview or a successful automated check alone is not proof of accessibility.
Why PDF Accessibility Matters
- Legal requirements — accessibility duties vary by jurisdiction, organization and service. Check the law, procurement terms and standards that apply to your document rather than assuming a universal PDF/UA mandate.
- Government requirements — US Section 508 and European requirements using EN 301 549 can apply to electronic documents within their defined scope. Applicable criteria and exceptions require a context-specific review.
- Education — students with visual impairments need accessible course materials, syllabi, and exam papers.
- Searchability — extractable text helps search and indexing. Tags provide semantic structure but are not a prerequisite for Google to index a PDF.
- Reflow — properly tagged PDFs can reflow text for mobile viewing and zoom without horizontal scrolling.
Tagged PDF vs Untagged PDF
The foundation of PDF accessibility is the tag tree — a logical structure that describes the document's content hierarchy.
| Aspect | Untagged PDF | Tagged PDF |
|---|---|---|
| What a screen reader sees | Text and structure inferred by the viewer; results vary | Headings, paragraphs, lists, tables, images with alt text |
| Reading order | Inferred from content and layout; may be incorrect | Defined by the tag tree; must be checked for correctness |
| Tables | Read as disconnected text fragments | Read as rows and columns with headers |
| Images | Meaning may be unavailable without suitable alternatives | Announced with alt text description |
| Reflow | Some viewers infer reflow, with variable results | Can support reflow when structure and viewer support are correct |
Essential PDF Tags
PDF tags map to familiar HTML-like semantics:
| PDF Tag | HTML Equivalent | Purpose |
|---|---|---|
| <Document> | <html> | Root element of the tag tree |
| <H1> — <H6> | <h1> — <h6> | Heading hierarchy for navigation |
| <P> | <p> | Paragraph text |
| <L>, <LI> | <ul>/<ol>, <li> | Lists and list items |
| <Table>, <TR>, <TH>, <TD> | Same names | Table structure with header cells |
| <Figure> | <figure> | Images and illustrations (holds alt text) |
| <Link> | <a> | Hyperlinks |
| <Form> | <form> | Interactive form containers |
| /Artifact marked content (outside the tag tree) | n/a | Decorative content to be ignored (headers, footers, page numbers, watermarks) |
Reading Order
PDFs are not like HTML — there's no inherent "top to bottom" reading order. Content can appear anywhere on the page in any byte order. The tag tree defines the logical reading sequence.
Common reading order problems:
- Multi-column layouts — a two-column page may read left column top to bottom, then right column, or it may zigzag between columns depending on how it was created.
- Headers and footers — page numbers and running titles should be marked as
/Artifactoutside the structure tree so screen readers skip them. - Sidebars and call-out boxes — must be placed in the correct position in the tag tree, not just visually on the page.
- Footnotes — should be linked to their reference points and positioned logically in the reading order.
Alt Text for Images
Every meaningful image needs alternative text. The rules:
- Describe the content, not the appearance — "Bar chart showing Q3 revenue up 15% to $2.3M" is useful. "Image of chart" is not.
- Be concise — aim for one or two sentences. For complex images like data charts, consider a longer description or a data table alternative.
- Skip decorative images — purely decorative borders, dividers, and background images should be marked as
/Artifactoutside the structure tree with no alt text. - Avoid "image of" or "picture of" — screen readers already announce it as an image. Saying "image of a dog" is redundant; just say "Golden retriever in a park".
Accessible Tables
Tables are one of the hardest elements to make accessible in PDF:
- Data tables need appropriate header semantics and associations so readers can relate cells correctly. Do not add misleading header tags to layout-only content; complex tables need additional manual checking.
- Use the
Scopeattribute on headers to specify whether they apply to the row or column. - Avoid merged cells when possible — they confuse screen readers. If you must merge, define the header relationships explicitly.
- Don't use tables for layout — only for tabular data. Layout tables are a major accessibility barrier.
Language Settings
The document language tells screen readers which pronunciation rules and dictionaries to use:
- Set the primary language in the document properties (e.g., "en-US" for American English).
- For multilingual documents, tag individual sections with their language (e.g., a French quote within an English document).
- Without language settings, screen readers may mispronounce every word using the wrong language model.
Accessible Forms
- Every form field must have a label that describes what to enter (not just placeholder text).
- Group related fields (e.g., "Address" fields) using
<Form>tags. - Provide tool tips for fields — screen readers announce these as instructions.
- Ensure tab order follows the logical reading order (top to bottom, left to right).
- Include error messages that are announced when validation fails.
PDF/UA Standard
PDF/UA (ISO 14289) is the international standard for accessible PDF. Key requirements:
- All meaningful content needs appropriate structure; decorative content is marked as artifacts outside the structure tree.
- Tags must be in a correct, meaningful order.
- All images have alt text or are marked as artifacts.
- Document language is set.
- Font information allows Unicode mapping (for text extraction and search).
- Bookmarks reflect the document's heading structure.
- Security settings don't block assistive technology access.
Testing Tools
| Tool | Cost | What It Checks |
|---|---|---|
| Adobe Acrobat Accessibility Check | Paid (Acrobat Pro) | Automated structural checks; reading order and alternative-text quality also require manual review |
| PAC (PDF Accessibility Checker) | Free | Automated PDF/UA checks and reports; human checks remain necessary |
| NVDA (Screen Reader) | Free | Real-world screen reader testing on Windows |
| VoiceOver | Free (built into macOS/iOS) | Real-world screen reader testing on Apple devices |
| veraPDF | Free | Machine-verifiable PDF/UA requirements; not a replacement for human checks |
Accessibility Checklist
- Tag tree with proper heading hierarchy (H1 → H2 → H3, no skipping)
- Alt text on all meaningful images
- Decorative elements marked as artifacts
- Correct reading order (inspect tags and test with assistive technology)
- Table headers defined with TH tags and scope
- Document language set
- Form fields labeled
- Tab order matches logical reading order
- Sufficient color contrast (4.5:1 for body text)
- Bookmarks match heading structure
- Tested with PAC and a screen reader