PDF/UA explained: PDF/UA-1, PDF/UA-2 and the Matterhorn Protocol
PDF/UA is the ISO standard for accessible PDF files. What PDF/UA-1 and PDF/UA-2 require, how Matterhorn and validators test them, and the identifier.
Last updated
PDF/UA (PDF for Universal Accessibility) is the ISO standard that defines what a PDF file must contain to be accessible to people using assistive technology. It's published as ISO 14289. Part 1, PDF/UA-1 (ISO 14289-1:2014), applies to PDF 1.7 files, and Part 2, PDF/UA-2 (ISO 14289-2:2024), applies to PDF 2.0 files. The Matterhorn Protocol, from the PDF Association, turns PDF/UA-1 into a list of failure conditions: some can be tested by software, and some need a person. Validators such as veraPDF and PAC test the software-checkable ones.
What PDF/UA requires
In plain English, a PDF/UA file must meet requirements in these areas:
- Structure. The file is tagged. All real content is tagged in a logical reading order, and everything else (page numbers, running headers, decoration) is marked as an artifact. Tags are standard types, or custom types mapped to standard ones, and they match what the content is: headings are tagged as headings, lists as lists, tables as tables.
- Text. Every font is embedded, and every character maps to Unicode, so text reads and copies correctly. The document's language is declared, along with any changes of language.
- Images and math. Figures and formulas have alternate text.
- Tables. Header cells are tagged as headers, and data cells can be connected to them.
- Metadata. The file has a title in its metadata, shows that title in the window bar, and declares PDF/UA conformance with an identifier.
- Interaction. Links, form fields and other annotations are tagged and described; the tab order follows the structure; dynamic XFA forms aren't used.
- Security. Encryption doesn't stop assistive technology from reading the content.
PDF/UA also points to WCAG for some content questions, such as conveying information through color or contrast. Our WCAG vs. PDF/UA guide explains where the two standards overlap and where they differ.
PDF/UA-1 vs. PDF/UA-2
| PDF/UA-1 | PDF/UA-2 | |
|---|---|---|
| Standard | ISO 14289-1 (first published in 2012; current edition 2014) | ISO 14289-2:2024 |
| Based on | PDF 1.7 (ISO 32000-1) | PDF 2.0 (ISO 32000-2) |
| Tag set | The PDF 1.7 standard structure types | The PDF 2.0 structure types, which add elements such as Title, Aside and FENote |
| Test checklist | The Matterhorn Protocol | Validator profiles such as veraPDF's |
| Tool support | Broad: most authoring and remediation tools, and every major validator | Growing: newer tools and validators |
The PDF Association also publishes Well-Tagged PDF (WTPDF) 1.0, a free specification for tagging PDF 2.0 files. Its accessibility conformance level is aligned with PDF/UA-2, so the two are often mentioned together.
Which should you target? For most organizations today, PDF/UA-1 is the practical choice: it's what most authoring tools produce and what PAC and most remediation workflows test. PDF/UA-2 matters as more tools move to PDF 2.0. Neither is named in the ADA Title II rule, which requires WCAG 2.1 AA.
The Matterhorn Protocol
The PDF/UA standard says what a conforming file must contain, but not how to test it. The Matterhorn Protocol, published free by the PDF Association, fills that gap for PDF/UA-1. It breaks the standard into 31 checkpoints, such as real content tagged, metadata, graphics, headings, tables, annotations and fonts, and 136 failure conditions. Each failure condition is labeled as either machine-checkable or requiring human judgment. In the current version, Matterhorn Protocol 1.1 (2021), 87 failure conditions can be checked by software, 47 usually need human judgment, and 2 have no specific test.
Examples:
| Failure condition (in plain English) | How it's tested |
|---|---|
| Content is neither tagged nor marked as an artifact | Machine |
| A figure has no alternate text | Machine |
| The alternate text doesn't describe the figure | Human |
| A font isn't embedded | Machine |
| Text that looks like a heading isn't tagged as one | Human |
| The reading order isn't logical | Human |
| The document's title is missing from the metadata | Machine |
| The PDF/UA identifier is missing | Machine |
This split is the most important thing to understand about PDF/UA checking. A validator can only test the machine-checkable conditions. "PAC passed" means the machine checks passed, not that the file conforms to PDF/UA. Conformance also needs a person to work through the human-judgment conditions.
Validators
| Tool | What it is | Good to know |
|---|---|---|
| veraPDF | Free, open-source validator from the PDF Association and the Open Preservation Foundation | Validates PDF/UA and PDF/A, and runs from the command line or a desktop app. Includoc uses it. |
| PAC (PDF Accessibility Checker) | Free Windows app, currently PAC 2026 | Tests machine-checkable PDF/UA (Matterhorn) conditions and many WCAG requirements, and has a screen reader and structure preview for the visual checks |
| Acrobat Pro's Accessibility Checker | Built into Adobe Acrobat Pro | Uses Adobe's own rules, so its results don't map one to one to PDF/UA |
| Includoc | Online checker | veraPDF validation plus our own checks, with each result mapped to Matterhorn, PAC and Acrobat wording |
Different validators sometimes disagree on edge cases, so specialists often use two. Whatever you use, record the tool and version in your report.
Common PDF/UA failures and what they mean
These are among the failures people search for most often. Each links to a page with step-by-step fixes.
| What the report says | What it means in plain English |
|---|---|
| Content is tagged or artifacted | Some text or graphics are neither tagged as content nor marked as decoration |
| Non-standard structure type is remapped | A custom tag, often named after a Word or InDesign style, isn't mapped to a standard tag |
| Figure tag problems | A Figure tag has no alt text, or holds content that isn't an image |
| Link annotations in Link tag | Clickable links aren't inside Link tags in the tag tree |
| Widget annotations in Form tag | Form fields aren't inside Form tags |
| Annotation has alt text | A comment, stamp or other annotation has no description |
| TH has associated cells | A table header cell isn't linked to the data cells it describes |
| Fonts are embedded | A font the text needs isn't included in the file |
| PDF/UA identifier missing | The metadata doesn't declare PDF/UA conformance |
PAC also reports some checks that go beyond the PDF/UA-1 failure conditions, such as figure has bbox, which helps figures display correctly when content is reflowed or magnified.
The PDF/UA identifier
The PDF/UA identifier is a small piece of metadata (in the file's XMP metadata) that declares "this file conforms to PDF/UA". For PDF/UA-1, it's the pdfuaid:part property with the value 1. PDF/UA-2 files use the value 2 and also record the revision year.
- PAC and veraPDF report a missing identifier as a failure, even though it doesn't change what anyone hears. See PDF/UA identifier missing.
- Only add it when the file actually conforms. Adding the identifier to a file that fails other requirements is a false claim, and it can mislead the next person who checks the file.
Is PDF/UA required by law?
Not directly, in the US. The ADA Title II and HHS Section 504 web rules require WCAG 2.1 Level AA, and they apply to the documents you post. When the Department of Justice wrote the Title II rule, it considered PDF/UA and declined to add it as a separate technical standard. But building to PDF/UA is a sound technical foundation for meeting WCAG in a PDF, and its free validators give you consistent, repeatable checks. Some contracts and procurement documents ask for PDF/UA explicitly, so check what yours require. Our ADA Title II guide covers the rule itself.
Fix it automatically
Includoc validates every PDF with veraPDF's PDF/UA-1 profile plus our own WCAG 2.1 AA checks, and maps each result to its Matterhorn checkpoint and the matching PAC and Acrobat message. When we fix a file, we tag all content or mark it as an artifact, embed the document title and language, fix the role map and tag annotations. We add the PDF/UA identifier only after every machine check passes, and we list the human-judgment items for you to confirm. Problems we can't safely fix after the fact, such as fonts that aren't embedded, are flagged with instructions to re-export from the source.
Upload your PDF to fix this automatically
Free check in seconds. Files are deleted within 24 hours.
Standards
| Standard or tool | Reference |
|---|---|
| PDF/UA-1 (ISO 14289-1) | Clause 5 |
| Matterhorn Protocol | Checkpoint 06-002 |
| PAC wording | PDF/UA identifier missing |
| Standard or tool | Reference |
|---|---|
| WCAG 2.1 | |
| PDF/UA-1 (ISO 14289-1) | Clause 7.1 |
| Matterhorn Protocol | Checkpoint 01-005 |
| PAC wording | Content is tagged or artifacted |
| Acrobat rule | Tagged content |
| Standard or tool | Reference |
|---|---|
| PDF/UA-1 (ISO 14289-1) | Clause 7.21.4.1 |
| PAC wording | Fonts are embedded |
Sources
- ISO: ISO 14289-1:2014, PDF/UA-1 (opens another website)
- ISO: ISO 14289-2:2024, PDF/UA-2 (opens another website)
- PDF Association: The Matterhorn Protocol 1.1 (opens another website) (PDF (opens another website))
- PDF Association: Well-Tagged PDF (WTPDF) 1.0 (opens another website)
- veraPDF: Project website (opens another website) and PDF/UA-1 validation rules (opens another website), including the identifier rule
- Federal Register, April 24, 2024: ADA Title II final rule (opens another website)
- PAC: PDF Accessibility Checker (opens another website) and its download and system requirements (opens another website)
- W3C: PDF Techniques for WCAG (opens another website)
Related
- WCAG vs. PDF/UA: which standard do your PDFs need?
WCAG 2.1 AA is what the ADA Title II rule requires; PDF/UA is the ISO standard for accessible PDF files. How they differ, overlap and map to each other.
- What is a tagged PDF? Tags, the tag tree and how to check them
A tagged PDF has hidden structure that tells screen readers what each part is. Learn the common tags, artifacts, role maps and how to view and fix tags.
- How to make a PDF accessible: a step-by-step guide
Make a PDF accessible from start to finish: fix the source file, export with tags, then check and fix tags, reading order, alt text, tables and forms.
- PDF/UA identifier missing
PAC reports “PDF/UA identifier missing” when a PDF’s metadata doesn’t claim PDF/UA. Why to add it last, only after the file passes, and how.
- Content is tagged or artifacted
PAC fails “Content is tagged or artifacted” when page content is neither tagged nor marked as decoration. What it means and how to fix it.
- Fonts are embedded
PAC fails “Fonts are embedded” when a PDF uses fonts it doesn’t include. Why PDF/UA requires embedded fonts, how to check, and how to fix it.