10 min read 19 October 2026

Tagged PDF: Structure Trees, Role Maps, and Reading Order

Tagged PDFs carry a logical structure tree that screen readers and reflow engines use to interpret document meaning. This guide explains the tag tree, standard structure element types, role maps, marked content sequences, and how tagging enables PDF/UA accessibility and document reflow.

Tagged PDF adds a layer of semantic structure to a PDF that is invisible in normal viewing but essential for accessibility and document reflow. Without tags, a PDF is a collection of drawing instructions — paint this text at these coordinates, draw this line here — with no information about what that content means. With tags, the same PDF carries a logical structure tree that identifies headings, paragraphs, lists, tables, figures, and other elements, enabling screen readers to read the document correctly, reflow engines to re-paginate it for smaller screens, and search engines to understand its organisation. Tagged PDF is the technical foundation of PDF/UA (ISO 14289) accessibility compliance. It is also required by Section 508, WCAG 2.1 PDF techniques, and increasingly by EU public sector accessibility legislation. This guide covers the structure tree, standard element types, role maps, marked content sequences, and how to create and validate tagged PDFs.

The structure tree and marked content sequences

Tagged PDF operates through the interaction of two parallel systems: The structure tree (stored in the document catalogue's /StructTreeRoot) is a tree of structure element objects. Each structure element has a type (e.g., /H1, /P, /Table), optional attributes, and children that are either other structure elements or content references. Marked content sequences are regions in page content streams delimited by the BMC (Begin Marked Content) or BDC (Begin Dictionary Marked Content) and EMC (End Marked Content) operators. Each sequence has a tag name and optionally an inline or referenced property dictionary. Structure elements in the structure tree reference marked content sequences via marked content reference (MCR) objects that identify the page and the /MCID (Marked Content Identifier) of the sequence. This two-part system means that: (1) The structure tree describes the logical document organisation; (2) The marked content sequences identify exactly which page content corresponds to each structure element; (3) Structure elements reference marked content sequences to connect the logical structure to the visual content. Object references — Structure elements can also reference annotation objects (form fields, links) directly via object references (OBJR), not marked content sequences. A Link annotation is referenced from a /Link structure element via OBJR.

Standard structure element types

The PDF specification defines a standard set of structure element types organised into several categories: Document-level grouping elements: - /Document — Root element of a complete document - /Part — A large-scale division of a document (part, chapter) - /Art — An article (self-contained composition) - /Sect — A section or chapter subdivision - /Div — A generic block-level grouping element - /Aside — Tangential content (sidebar, pull quote) — new in PDF 2.0 / PDF/UA-2 Block-level elements: - /H, /H1–/H6 — Headings (H for generic heading, H1–H6 for levelled headings) - /P — Paragraph - /L — List container; /LI — List item; /Lbl — Label; /LBody — Body - /BlockQuote — Block quotation - /Caption — Caption for a figure or table - /Index — Index - /TOC / /TOCI — Table of contents / Table of contents item - /FENote — Footnote or endnote — new in PDF 2.0 / PDF/UA-2 Inline-level elements: - /Span — Generic inline text element (used for language changes, abbreviations) - /Quote — Inline quotation - /Note — Footnote reference marker - /Reference — Cross-reference in text - /BibEntry — Bibliography entry Table elements: - /Table — Table - /TR — Table row - /TH — Table header cell - /TD — Table data cell - /THead, /TBody, /TFoot — Table head/body/foot groups Illustration elements: - /Figure — Figure, photograph, or graphic (must have /Alt text) - /Formula — Mathematical formula - /Form — Form widget (AcroForm field)

Role maps and custom element types

The standard element types cover most common document structures, but specialised documents may need custom element types. PDF's role map mechanism allows this. A role map (/RoleMap in the structure tree root) is a dictionary that maps custom element type names to standard element types. For example, a publisher might use /Chapter as their structure element type, mapped to /Sect in the role map. PDF viewers and accessibility tools that encounter /Chapter resolve it to its mapped standard type /Sect and treat it accordingly. Role maps are important for: - Legacy documents that use non-standard tag names from older tagging tools - Domain-specific documents that need precise semantics not captured by the standard set - PDF/UA compliance — every structure element type in a PDF/UA document must either be a standard type or be mapped via the role map to a standard type The Matterhorn Protocol (published by the PDF Association) is the definitive test suite for PDF/UA-1 conformance. It defines 136 checkpoints organised into categories, many of which relate directly to structure tree correctness, role map completeness, and marked content integrity. PAC (PDF Accessibility Checker) implements Matterhorn Protocol checks.

Creating tagged PDFs from authoring tools

Microsoft Word to PDF — Word's built-in PDF export (Save As > PDF) creates a tagged PDF when the "Document structure tags for accessibility" option is enabled (checked by default in current Word for Windows). Heading styles (Heading 1, Heading 2) export as H1, H2 tags. Properly styled lists export as L/LI/LBody structures. Images with alt text export with /Alt attributes. The tag quality is generally good for simple documents but may require review for complex layouts. Adobe InDesign to PDF — InDesign's PDF export with "Create Tagged PDF" checked produces tagged output. InDesign gives direct control over the tag hierarchy via the Articles panel and Object Export Options. For print-quality publications, InDesign produces the highest-quality tagged PDF of any mainstream authoring tool. Adobe Acrobat Pro — manual tagging and remediation — Tools > Accessibility provides the Reading Order tool (for quick initial tagging), the Tags panel (for detailed structure tree editing), and the accessibility checker. The Add Tags to Document feature runs an auto-tagger that requires human review. For full remediation of complex documents, the Tags panel is the primary tool. PDF libraries — iText (commercial) has full tagged PDF support. Apache PDFBox can create tagged PDFs via its PDDocument and PDStructureElement APIs. Typesetting systems like LaTeX (with the tagpdf package) and CSS Paged Media implementations (Prince, Antenna House) produce well-tagged output.

Frequently asked questions

How can I check if a PDF is tagged?

Several methods: (1) Adobe Acrobat: File > Properties > Description tab — look for "Tagged PDF: Yes". (2) Acrobat accessibility checker: Tools > Accessibility > Full Check — the first check is whether the document is tagged. (3) PAC (PDF Accessibility Checker) — a free tool that provides a detailed tagged PDF validation report including structure tree visualisation. (4) Command-line with qpdf: qpdf --qdf input.pdf output.qdf — then search the output for /MarkInfo in the document catalogue; /Marked true indicates a tagged PDF. (5) In the PDF source: look for /MarkInfo << /Marked true >> in the document catalogue dictionary. A PDF without this entry is untagged. Note: a PDF can be "tagged" (has /MarkInfo /Marked true) but have poor tag quality — tagging presence is a prerequisite for accessibility, not a guarantee of it.

What is the reading order in a tagged PDF and why does it matter?

Reading order defines the sequence in which a screen reader announces content to a visually impaired user. In a tagged PDF, reading order is determined by the order of structure elements in the structure tree (not the visual layout). A two-column PDF, for example, might have its content stream arranged left-then-right across both columns for rendering efficiency, but the structure tree should have all the left column content before all the right column content — matching the intended reading sequence. When reading order is wrong, a screen reader reads columns out of order or announces headings after the content they precede. Acrobat Pro's Reading Order tool lets you visually reorder structure tree elements. The order of elements in the Tags panel is the reading order; dragging elements reorders them.

What alternative text is required for images in a tagged PDF?

Every /Figure structure element in a tagged PDF must have either: (a) an /Alt attribute containing a text description of the image, or (b) /ActualText if the image contains text (e.g., a scanned word). The alternative text must convey the content and purpose of the image — not just "image" or a filename. For decorative images that add no informational value, the /Figure element should be marked as an artifact (using /Artifact marked content) rather than a /Figure structure element — this tells screen readers to skip it. In Acrobat Pro: select a Figure in the Tags panel, right-click > Properties > to add /Alt text. In Word: right-click an image > Edit Alt Text. WCAG 2.1 Success Criterion 1.1.1 (Non-text Content) applies to all images in accessible documents including PDFs.

Try it yourself

Everything in this article can be done directly in the free PDFTheGuide editor — no account required.

Edit PDF Annotate PDF Open editor