Standard
PDF/UA-1 (ISO 14289-1): what it actually requires
The file-format standard for accessible PDF. What each clause asks for, how conformance is tested through the Matterhorn Protocol
ISO — ISO 14289-1:2014 · technical standard: ISO 32000-1 (PDF 1.7) with accessibility requirements
What the standard is for
PDF/UA-1 answers one question: what must be in a PDF file so that assistive technology can present it faithfully?
It is deliberately narrow. It says nothing about whether your prose is clear, whether your colours have enough contrast, or whether your form is sensibly designed. It says that the file must carry a logical structure, that structure must reflect the reading order, every piece of content must be either tagged or declared an artifact, text must be mappable to Unicode, and a natural language must be determinable.
Those are file-format facts, and they are machine-checkable in a way that “is this alt text any good” is not.
The clauses, in the order you meet them
7.1 — General. The foundation. Content is either real content in the structure tree or an explicit artifact, with no third option. /MarkInfo /Marked must be true, /Suspects must be false, and every non-standard tag must role-map to a standard type. The untagged-content condition →
7.2 — Text. Reading order must be logical. Every character code must be mappable to Unicode. A natural language must be determinable for page content, for Alt and ActualText, for outline entries, for annotation contents and for form field tooltips. Reading order → · Language →
7.3 — Graphics. Meaningful graphics are Figure elements with /Alt or /ActualText; decorative ones are artifacts. Captions belong in Caption. Figure alt text →
7.4 — Headings. Either the strong-structure model (H levelled by nesting) or the weak one (H1–H6), never both. Numbered levels must start at H1 and must not skip going down.
7.5 — Tables. Header cells must declare /Scope, or the table must use explicit /Headers and /ID references. TH scope →
7.6 to 7.12 — lists, mathematics, page headers and footers, notes and references, optional content, embedded files, article threads. Each with its own small set of requirements.
7.16 — Security. If the file is encrypted, bit 10 of the permissions must allow extraction for accessibility. One bit, and it can render a perfect document unusable.
7.18 — Annotations. Links inside Link elements, widgets inside Form elements, tab order following structure, and an accessible description on everything.
7.21 — Fonts. Thirty of the protocol’s failure conditions live here. Fonts must be embedded, encodings must be consistent, and a ToUnicode CMap is required unless a specific exemption applies. ToUnicode →
What a conformance claim is worth
The PDF/UA identifier in the XMP metadata is an assertion by whoever wrote the file. Nothing validates it, and plenty of files carry it while failing.
This is why Taggart will not write it on request. It writes it only after re-validating the output it just produced and finding no blockers, and after a person has confirmed the conditions Taggart does not check — and when it declines, the report says exactly why:
conformance was requested but re-validation found 3 blocker(s); the PDF/UA identifier was not written
Even then, what Taggart verified is the machine-checkable conditions it checks; the rest of the claim rests on that confirmation. The 47 requiring human judgement are recorded as decisions with an actor and a timestamp, and the report’s conformance statement distinguishes between “every machine check passes” and “every machine check passes and a human has decided the rest”.
Frequently asked questions
Is PDF/UA the same as WCAG?
No. PDF/UA-1 is a file-format standard: it says what a conforming PDF file must contain. WCAG 2.1 is a content standard that applies across formats. They overlap heavily — most PDF/UA requirements map to a WCAG success criterion — but each has requirements the other does not. A document can conform to PDF/UA-1 and still fail WCAG on colour contrast, for instance.
How is PDF/UA-1 conformance tested?
Through the Matterhorn Protocol, published by the PDF Association: 31 checkpoints comprising 136 failure conditions, of which 87 can be decided by software and 47 require human judgement. Two have no specific test at all. A tool that claims to verify PDF/UA-1 fully is claiming to automate the 47, which it cannot.
What does the PDF/UA identifier in the metadata mean?
It is a claim, not a verification. Writing pdfuaid:part 1 into the XMP packet asserts conformance; nothing checks it. That is why Taggart refuses to write it unless its own re-validation of the output finds no blockers, and says so explicitly when it declines.
Is there a PDF/UA-2?
Yes — ISO 14289-2, built on PDF 2.0. It is the direction of travel, and Taggart models its checks against a rule file rather than hardcoding clauses so a UA-2 profile can be added without rewriting the engine.
Related guides
- How to make a PDF accessibleThe whole job end to end: what has to be true, what order to do it in, what is genuinely mechanical, and where a person has to decide — with real timings.
- What is a tagged PDF?A PDF page is drawing instructions with no idea what it draws. Tags are the structure that says. Why they exist, and why tagged does not mean accessible.
- How to check if a PDF is accessible: four free checkersCheck PDF accessibility free, online or on your own machine: four checkers against PDF/UA and WCAG, what each catches, and the failures none of them see.
- PDF/UA vs WCAG: which one applies to your documentsOne is a format standard a validator can test. The other is what your legal obligation cites. They overlap heavily and neither one contains the other.
Related failure conditions
- 01-005 Content is neither marked as Artifact nor tagged as real contentEvery text-showing, image-painting and path-painting operator on a page must sit either inside a marked-content sequence that maps to the structure tree, or inside an artifact sequence. Content that is in neither is orphaned.
- 09-001 Tags are not in logical reading orderThe order of elements in the structure tree does not match the order a human would read the page. This is the single most common substantive defect in tagged PDFs and the one Acrobat's own tooling handles worst.
- 11-001 Natural language for text in page content cannot be determinedNo /Lang is in effect for some text: neither on the catalog, nor on an enclosing structure element, nor on a marked-content sequence.
- 13-004 Figure tag alternative or replacement text missingA Figure element has neither /Alt nor /ActualText. Every Figure must have one or the other; if the image is decorative it should be an artifact instead of a Figure.
- 31-027 A font dictionary does not contain the ToUnicode entry and no exemption appliesA ToUnicode CMap is required unless the font uses MacRoman/MacExpert/WinAnsi encoding, or is a Type 1/Type 3 whose glyph names are all in the Adobe Glyph List or the Symbol set, or is a Type 0 whose descendant uses a standard Adobe character collection, or is a non-symbolic TrueType.
Check your own file. Taggart's validator is free and unlimited — no page cap, no watermark, and no account at all up to 60 pages.
Primary source: https://www.iso.org/standard/64599.html. This page is a working summary, not legal advice.
Last updated .