How to Fix the Reading Order in a PDF to Make It Screen Reader Friendly

How to Fix the Reading Order in a PDF to Make It Screen Reader Friendly

Screen readers follow a PDF's tag tree, not its visual layout. How to find the reading order that matters, fix it at the source or in the PDF, and test.

A screen reader does not read a PDF the way you look at it. It walks the document's tag tree in order, and that order has nothing to do with where things sit on the page. This is why a clean two-column layout can be read as a single scrambled stream, alternating between columns mid-sentence.

The tag tree is the reading order

The structure tree is a sequence: heading, paragraph, paragraph, figure, table, and so on. Assistive technology follows it from top to bottom. Whatever order it holds is the reading order, even if it makes no sense visually.

Content is usually added to the tree in the order the exporter emitted it, which often follows the drawing order in the file rather than the layout. A sidebar drawn last is read last, after the main text, even though a sighted reader meets it halfway down page one.

Two panels in Acrobat, and only one of them matters

This confuses people constantly, and getting it wrong means hours of work with no effect.

The Order panel (sometimes surfaced by the Reading Order tool) shows a page-by-page sequence and is convenient for coarse reordering.

The Tags panel shows the actual structure tree. This is what assistive technology reads.

They are related but not identical, and on complex documents changes made in one do not always produce what you expect in the other. When you need certainty — and for anything being audited, you do — work in the Tags panel and verify there.

Fix the source rather than the PDF, where you can

Repairing a tag tree by hand is slow, and it has to be redone every time the document is revised. Fixing the source is faster and it lasts.

In Word, use real heading styles rather than manually enlarged bold text, put content in the main story rather than floating text boxes, and export with "Document structure tags for accessibility" enabled. Text boxes are a common culprit: they frequently land at the end of the reading order regardless of position.

In InDesign, use the Articles panel to define the reading order explicitly, and export with Create Tagged PDF enabled. Without an Articles definition, InDesign's order can be close to arbitrary on a complex layout.

In both cases the export setting is easy to miss and it is the whole game — without it you get an untagged PDF and no reading order at all.

Fixing it in the PDF

When the source is gone, the work is in the Tags panel of a tool that can edit structure — Acrobat Pro being the common one:

  1. Check it is tagged. An untagged document needs tagging before anything can be reordered. Autotagging gets you a starting point, not a finished job.
  2. Remove page furniture from the tree. Running headers, footers and page numbers should be artifacts, not content. Otherwise the page number is announced mid-sentence at every page break.
  3. Drag elements into logical order. Follow how a person reads: heading, then its text, then the figure it refers to.
  4. Check tables separately. A table's reading order depends on its row and cell structure, and a table built from drawn lines with no table tag will be read as loose text.
  5. Test with an actual screen reader. NVDA on Windows is free. Nothing substitutes for hearing the document read.

Be clear about the boundary here: this site's editor changes page content — text, images, layout — on files up to 10 MB and 100 pages. It does not edit the tag tree, so it is not the tool for reordering structure.

The layouts that break most often

Multi-column text read across columns instead of down them. Usually the columns were never separated in the tree.

Sidebars and pull quotes interrupting the main text mid-paragraph.

Figure captions separated from their figure, so an unexplained caption arrives somewhere unrelated.

Tables used for layout rather than data, read cell by cell as nonsense.

Scanned pages with an OCR text layer, where the recognised text order follows the scan's internal layout analysis and can be close to random.

Frequently asked questions

How do I know what the reading order is? Read the Tags panel top to bottom, or listen with a screen reader. The visual layout tells you nothing about it.

Why is my page number read in the middle of a sentence? It is tagged as content instead of being marked as an artifact.

Does autotagging fix reading order? It produces a plausible starting point and frequently gets complex layouts wrong. Always review the result.

Are the Order panel and the Tags panel the same? No. The Tags panel is the structure assistive technology reads. Verify there.

Can I fix reading order for free? Not easily in the PDF. Fixing the source and re-exporting with tagging enabled is the realistic free route.

Usama Ramzan
Written byUsama RamzanFounder, Online PDF Edits

Usama Ramzan is the founder of Online PDF Edits, a browser-based PDF editor built to change text, images, and tables in existing PDFs without breaking their fonts, spacing, or multi-page layout. He writes about practical PDF editing, document workflows, and the engineering behind layout-safe editing.

Recommended reading

View all articles →