Accessibility Compliance
How to Make PDFs Accessible (PDF/UA)
Short answer
PDF/UA is ISO 14289: PDF/UA-1 (ISO 14289-1) for PDF 1.7 documents and PDF/UA-2 (ISO 14289-2:2024) for the newer PDF 2.0 format. Both define accessible PDFs through a correct tag tree, reading order, alt text, table headers, form labels and language tagging - requirements a visually normal-looking PDF can completely fail. Scanned documents without OCR and proper tagging are typically the worst offenders and the biggest remediation projects.
A PDF that looks perfectly normal on screen can be entirely unusable to someone using a screen reader - no reading order, no way to know what a scanned image says, form fields with no labels. PDF/UA - ISO 14289: PDF/UA-1 (ISO 14289-1) for PDF 1.7, PDF/UA-2 (ISO 14289-2:2024) for PDF 2.0 - is the international standard that defines what makes a PDF genuinely accessible, and it's increasingly what regulators and procurement teams check documents against. Here's what it requires in practice.
Why a visually fine PDF can still fail completely
Most inaccessible PDFs share one root cause: they were created by exporting from a design tool or scanning a paper document, with no attention to the document's underlying structure. A screen reader doesn't see the visual layout - it reads the document's tag tree, and if that tree doesn't exist or is wrong, the reading experience ranges from confusing to completely blocked.
This is why a PDF can pass a visual design review, a brand-compliance check and even basic print QA while being completely unusable to a blind user - none of those reviews look at the underlying tag structure at all, because it's invisible unless you specifically inspect it.
Tools worth knowing, and their limits
PAC (PDF Accessibility Checker) is the most widely used free validation tool and a reasonable first-pass check, but like any automated accessibility tool it validates structural compliance (is there a tag tree, are headers associated) without confirming the tags are semantically correct for the actual content - a table tagged as a table but with headers associated to the wrong columns will often pass automated validation while still being confusing to an actual screen-reader user.
What PDF/UA actually requires
- A complete, correctly ordered tag tree reflecting the logical reading order - headings, paragraphs, lists and tables all properly tagged, not just visually styled.
- Alt text for every meaningful image, and tags marking purely decorative images as artifacts so screen readers skip them.
- Table headers correctly associated with data cells, especially for complex or multi-level tables.
- Form fields with programmatic labels, not just visually adjacent text.
- A specified document language, with language changes marked where content switches mid-document.
- Correct bookmarks and a meaningful document title in the metadata, not a filename.
Scanned documents are usually the worst offenders
A scanned page is just an image to a screen reader, with zero text content, unless OCR has been run and the result properly tagged. Bulk-scanned document archives - old statements, historical forms - are almost always a bigger remediation project than born-digital PDFs.
Remediation approaches that actually scale
- Fix the template or export settings at the source, for PDFs generated from a CMS or document-generation system, so every future document is accessible by default.
- Batch remediation tooling plus manual QA for existing document archives, rather than fully manual one-by-one fixes.
- A PAC (PDF Accessibility Checker) or similar validation pass plus actual screen-reader testing, since automated PDF/UA checkers miss issues the same way web accessibility scanners do.
Prioritizing a large document archive
Most organizations with an accessibility obligation don't have one PDF to fix - they have thousands, accumulated over years. Trying to remediate everything at once is rarely realistic. A practical approach prioritizes by actual usage and legal exposure: documents linked from your website's main navigation, frequently downloaded forms, and anything a regulator or auditor is likely to specifically request, before working through a long tail of rarely-accessed historical documents.
Fixing the source instead of the symptom
If your organization generates PDFs programmatically - invoices, statements, policy documents - fixing the generation template once is worth more than remediating a thousand already-generated PDFs one at a time, because every future document inherits the fix automatically. This is usually the highest-leverage single change available: identify the two or three templates responsible for the majority of your PDF volume and fix those first.
Need help with this in practice?
A5 Cyber Nexus audits, remediates and documents conformance against these frameworks.
Frequently asked questions
Does making a PDF accessible mean just adding alt text?
No - alt text is one piece. Reading order, table structure, form labels, language tagging and bookmarks all matter, and automated tools that only check for alt text miss most real issues.
Should we just convert our PDFs to accessible HTML instead?
Sometimes that's the better answer, especially for frequently-updated or complex documents - but for legally or financially significant documents (contracts, statements) that need to retain their exact formatting, a properly tagged PDF/UA document is usually the right approach.
How do we prioritize which PDFs to fix first in a large archive?
By actual usage and legal exposure - documents linked from main navigation, frequently downloaded forms, and anything likely to be specifically requested by a regulator or auditor, before less-accessed historical documents.
Sources
Make it accessible.
Make it attack-ready.
Prove both before launch.
Bring us your app, audit deadline, or security concern. We'll map the fastest path to WCAG conformance, VAPT coverage, and regulator-ready evidence.
