Is my PDF accessible to screen reader users?
This check looks for five basics: tags, a title, a language, real text and bookmarks. Each one it finds adds a point, so the score runs from 0 to 5. A scanned page we tested scored 0 out of 5. A 5 means the basics are there, not that the file passes a full accessibility review.
How can I tell if a PDF is tagged?
Upload it and look at the Tagged PDF line. It passes only when the file says it is tagged and its tag tree holds at least one real element. One of our test files claimed to be tagged but had an empty tag tree, so it fails that signal. An untagged file needs a new export with tagged output turned on.
Why does a scanned PDF score 0 out of 5?
A scan is a picture of text. There is nothing a screen reader can read aloud, and scanners rarely add a title, a language, tags or bookmarks. Our image-only test page scored 0 out of 5. Run the file through OCR first, then tag the result before you check it again.
One point per signal
The score is a plain count of five yes-or-no signals, shown as Signals Present out of 5. Each signal is worth one point and nothing is weighted. 4 or 5 is shown in green, 2 or 3 in amber and 0 or 1 in red. The colours describe how many basic signals were found, not how usable the document is.
| Signal | Counted when | Read with |
|---|---|---|
| Tagged PDF | pdfinfo reports Tagged: yes, which reflects the catalog entry /MarkInfo << /Marked true >>, and the structure tree root holds at least one element in /K |
pdfinfo, then the PDF parser |
| Document Title | The Info dictionary holds a non-empty Title |
pdfinfo, then the PDF parser as fallback |
| Document Language | The document catalog has a non-empty /Lang entry, for example en-GB |
PDF parser |
| Text Content | More than 50 characters of text can be extracted from the whole file | PDF parser |
| Bookmarks / Outline | The catalog's /Outlines dictionary has a /First item, so at least one bookmark exists |
PDF parser |
Three measured files
The third file is the warning. It declares itself tagged and has an outline, but its structure tree contains no tags and its outline contains no bookmarks. Until 12 October 2026 the checker looked only for the presence of those entries and scored this file 5/5; it now requires at least one structure element and one bookmark, so the file scores 3/5. Even a 5/5 result only tells you the basics are present. Whether the tags describe headings, lists, tables and figures in the right reading order still has to be reviewed with a full checker or a screen reader.
Scores from synthetic test files
Text-only export with a title, no tags, no language, no outline Title: pass ("Quarterly report"). Text: pass. Tagged, language, bookmarks: missing. Score: 2/5 Image-only page, as a scanner would produce No title, no language, no tags, no outline, no extractable text. Score: 0/5 File with a MarkInfo flag, an empty structure tree, /Lang en-GB, a title, a paragraph of text and an outline containing zero bookmarks Title, language, text: pass. Tagged, bookmarks: missing. Score: 3/5
What each missing signal means for a reader
- Not tagged: a screen reader has no structure to follow and falls back to guessing reading order from the page layout, which often fails on columns, tables and footnotes. This is the hardest gap to fix after export; re-export from the source with tagged PDF output turned on.
- No title: assistive technology announces the file name instead, such as
scan_0042_final.pdf. A title is the quickest fix and can be added with a metadata editor. - No language: a screen reader uses its default voice and pronunciation rules, so an English document read with Lithuanian rules, or the reverse, becomes hard to follow. The language is set in the authoring application or with a full PDF editor.
- No text: the pages are images. Nothing can be read aloud, searched or reflowed until the file is run through OCR, and OCR output still needs tagging.
- No outline: long documents become tedious to move through. For short one or two page files a missing outline matters little, which is why it is shown in amber rather than red.
Details that change the result
- The title must be in the Info dictionary. A title stored only in XMP metadata is not counted.
- The 50-character threshold applies to the whole document. A scanned report with a short typed cover line can pass the text signal while every other page remains an image.
- The bookmarks signal needs at least one bookmark. An outline entry with zero items does not count, as the measured file shows.
- If the PDF parser cannot read the file, the tagged signal falls back to the pdfinfo flag alone, without the structure-tree check.
- Language is read from the document catalog only. Language tags set on individual passages are not examined.
What a full accessibility review covers that this check does not
PDF/UA and WCAG reviews cover much more than five signals: alternative text for figures, table header cells, a logical reading order, colour contrast, labelled form fields, artifacts marked so decorative content is skipped, and a PDF/UA identifier in the metadata. None of these are tested here. For a formal audit use a dedicated tool such as veraPDF in PDF/UA mode or PAC, and test with an actual screen reader.
Use this score as a fast triage step: it quickly separates files that were exported with no accessibility support at all from those worth a deeper review.
Last updated . Detection rules are taken from the checker's source code. The three sample scores were produced on 12 October 2026 by running the same service locally, with Poppler pdfinfo 26.10.0, on synthetic test PDFs we built for the purpose.