PDF to Word: conversion settings and limits
PDF to Word creates an editable Word document (DOCX) from the text in a searchable PDF. Files are processed in the browser on your device and are not uploaded.
Convert a PDF
- Open PDF to Word and select the PDF. To convert several files, select up to 50 PDFs; each becomes its own Word document, downloaded together as a ZIP.
- Leave Reconstruct paragraphs and basic formatting on for letters and reports, or turn it off to keep the extracted lines as they are.
- Turn on Attempt aligned table reconstruction if the PDF contains simple tables.
- Select Convert PDF to Word, then Download Word document.
Changing a setting or the file clears the result. You can cancel a conversion while it runs, and a conversion that takes longer than 90 seconds stops; split the PDF and try again.
Paragraphs and formatting
With paragraph reconstruction on, the tool groups text into lines and uses spacing, indentation and relative font size to work out paragraphs and headings. It keeps bold, italic and font sizes where the PDF provides them. Original typefaces are replaced with Arial. The result is an editable approximation, not an exact copy of the page.
Each page gets a layout status and any warnings. Large gaps, overlapping text, shifted blocks of text, rotated, vertical or right-to-left text, or inconsistent layout data make that page fall back to extracted lines. These checks do not catch every complex layout, so check the reading order against the original.
Tables
Attempt aligned table reconstruction is off by default. When it is on, simple whole-page grids become editable Word tables if at least three rows share two to twelve consistently placed columns. Missing or shifted cells, overlapping text, complex layouts and cells with several lines stay as text. Table boundaries are inferred, so check that a page of text in columns has not been turned into a table by mistake.
What is not converted
Images, complex tables, merged cells, columns, the original page layout, links, forms and annotations are not reconstructed. Each PDF page starts a new page in Word on A4 paper, but text reflow can add more pages. Hyphenated line endings are not joined automatically. Scanned pages need OCR first: pages without text get a placeholder.
The tool reads the source PDF, not your edits in the editor. Hidden or previously covered text can be extracted, so this is not a redacted export. Previewing the extracted pages is optional; compare the document with the source before you share it.