PDF Accessibility Checker
A PDF accessibility checker reads the structural signals that screen readers and other assistive technology depend on: whether the document is tagged, whether it declares its language and title, whether it carries a structure tree with real headings, whether its images have alternative text, and whether it has bookmarks. Drop a PDF below and this page checks those six signals locally, in your browser. Be clear about what this is from the start: it is a light structural check, not PDF/UA validation and not an accessibility certification — a fast smoke test that tells you which basics are present and which are missing, before you reach for full validation.
Last updated:
Or drop the file anywhere in this box. The file is read on this device and never uploaded. Limit: 20 MB.
Document facts
Accessibility checks
What this page is — and is not
This page performs a light structural check only. It is not a PDF/UA validation and it is not an accessibility certification: it cannot judge reading order quality, colour contrast, whether alt text is actually meaningful, or whether a screen-reader user can complete a task with the document. A file that passes all six checks here can still be hard to use, and a file that fails may need only small fixes. For full PDF/UA validation, use the free desktop validator veraPDF, and for remediation use a professional PDF accessibility tool. The companion article, your PDFs are part of your website’s accessibility, explains each check and its limits, and the Website Health Check covers the pages around the PDF.
Why these six signals matter
A PDF is not a picture of a document — at its best it is a structured file that software can navigate. Tags are the invisible structure that tells a screen reader what is a heading, a paragraph, a list or a table cell; an untagged PDF is often read as one undifferentiated stream, in whatever order the content happens to be stored. A declared document language lets speech synthesis pronounce the text with the right voice and rules instead of guessing. A document title is what assistive technology announces first and what a browser tab shows, which is why a file named by its export default — or with no title at all — is a small but real barrier. A structure tree with headings gives readers a way to skim and jump, the way sighted readers use a table of contents. Alternative text for images decides whether a chart, diagram or photograph exists at all for a blind reader. And bookmarks are the long-document navigation that makes a 90-page report usable instead of a scroll. None of these six guarantees an accessible document, but missing any of them is a concrete, fixable defect.
The checker itself is a hand-rolled parser that runs entirely in this page. It recovers the document catalog, the information dictionary and the structure tree from classic cross-reference tables, cross-reference streams and compressed object streams, inflating FlateDecode streams locally. When a structure cannot be recovered — a damaged file, an unusual producer, an arrangement the light parser does not follow — the affected check is reported as could not determine rather than guessed at, because a checker that invents verdicts is worse than one that admits its limits. Encrypted documents are reported as encrypted; their protected contents are not read. Everything happens on your device: analysing a file makes no network request, so internal reports, draft policies and personal documents stay where they are.
Frequently asked questions
Is my PDF uploaded anywhere?
No. The file is opened and parsed in this page, on your device, and nothing is uploaded, transmitted or stored. Organisations routinely publish PDFs that contain internal or personal information, so the checker was built to need no server at all: once the page has loaded, it works offline.
If a PDF passes all six checks, is it accessible?
Not necessarily — and this page will not claim otherwise. Passing means the six structural basics are present: tags, language, title, a structure tree, image alt coverage at the count level, and bookmarks. It says nothing about reading order, contrast, link text, table headers or whether the alt text says anything useful. Treat a full pass as “the smoke test is green”, then validate properly with veraPDF and, for important documents, test with a screen reader.
What does “tagged PDF” mean?
Tagging adds a hidden structure tree to the PDF that marks each piece of content — heading, paragraph, list, figure, table cell — and the order it should be read in. Word processors and layout tools can export tagged PDF when the option is enabled (in Microsoft Word, for example, the accessibility/tagging option in Save As / Export); scans and many quick exports produce untagged files. Tagging is the foundation the other checks build on: without a structure tree there are no real headings and nowhere to attach alt text.
How do I fix a failed check?
Fix at the source, not in the PDF. Turn on tagged-PDF export in your authoring tool, set the document language and title in the source file’s properties, use real heading styles instead of large bold text, write alt text for informative images in the source, and regenerate bookmarks from headings. Then re-export and check again. Editing a finished PDF with a remediation tool is possible, but every re-export from an unfixed source recreates the same failures — the article’s order of fixes walks through the sequence.