Repair an existing PDF
Triage a legacy PDF, locate its source, repair OCR, tags, reading order, links, and fields, then document human verification and remaining risks.
Triage before touching the file
Make a working copy and record the filename, date, owner, source application if known, and why the document is still needed. Check whether text can be selected, whether the file is tagged, whether links and form fields exist, and whether the PDF is a scan. Then look for an editable Word, presentation, spreadsheet, or page-layout source. Repairing that source and exporting a fresh tagged PDF is usually more durable than repeating Acrobat touchups every edition.
Prioritize files that people must use to complete a task, such as a public form, a meeting agenda, a safety instruction, or a report with a deadline. Do not remove content or alter a signed record merely to make the page simpler. If the file is a record, preserve the original and create a clearly identified accessible copy according to the site's records process.
Repair in a logical sequence
- Text: If it is a scanned PDF, run OCR and proofread the text against the page. OCR makes words selectable; it does not add headings, table relationships, correct language, or meaningful alt text.
- Document settings: Set the title, primary language, and a security setting that does not prevent assistive technology from reading.
- Structure: Add or correct tags, then inspect reading order. Mark decorative content as artifacts and apply heading, list, table, figure, and paragraph roles.
- Meaning: Add concise alt text, create bookmarks from headings for long files, and repair link names and destinations.
- Interaction: Label every form field, set required and input information, include fields in the tag tree, and make tab order follow the instructions.
Example
An image-only meeting agenda may need OCR, a title and language, heading tags for agenda sections, a corrected two-column reading order, bookmarks, link annotations, and a manual check that dates and motions are read in order. OCR alone does not complete that repair.
Verify and log the repair
Run Acrobat Pro's Accessibility Checker or Full Check after structural edits. It can identify missing tags, alternate text, titles, language, tab order, and other likely failures, but a passing panel is not proof that the document is usable or conforms to WCAG 2.2 AA. Manually inspect every page type and every task the reader must complete.
- Read the Tags tree and a representative page with a screen reader.
- Use the keyboard to reach every link and field, and verify visible focus and order.
- Use reflow or zoom, copy a passage, and compare extracted text with the visual page.
- Check tables cell by cell, including header announcements and blank or merged cells.
Log the page numbers changed, the source and tool versions, checker results, manual tests, unresolved limitations, and the next review date. Save the repaired file separately from the original and state which file should be published.
Escalate when repair is unsafe or ambiguous
Use a specialist for complex tables, mathematical notation, charts whose meaning is not captured by a short description, forms with conditional logic, damaged character encoding, or a document where the source and record requirements conflict. Include the original and working copy, a page-by-page defect log, and the user task. If a reliable repair cannot be completed before publication, provide the clearest accessible alternative available and record the limitation for follow-up; do not describe the PDF as fully conforming.
Before you publish
Check each item as you review. These marks last for this page visit only.
Sources and standards
Related WCAG criteria: 1.1.1 Non-text Content; 1.3.1 Info and Relationships; 1.3.2 Meaningful Sequence; 2.4.3 Focus Order; 2.4.4 Link Purpose (In Context); 2.4.6 Headings and Labels; 4.1.2 Name, Role, Value. This is a task-specific reference, not a complete conformance checklist.
- Adobe: Reading Order tool for PDFs
- Adobe: Create and verify PDF accessibility with Acrobat Pro
- W3C WAI: PDF7, performing OCR on a scanned PDF document
- W3C WAI: PDF3, ensuring correct tab and reading order in PDF documents
- W3C WAI: PDF20, using Acrobat Pro Table Editor to repair mistagged tables
Guidance reviewed September 12, 2026. Tool interfaces may vary by version.