How to Convert PDF to Word: 4 Methods, OCR and What Breaks

Adobe designed PDF in 1993 so a document would print the same on any machine, and the format became the international standard ISO 32000-1 in 2008. To keep that promise, a PDF page is a list of drawing instructions: place these glyphs at these coordinates in this font, draw a line here, put an image there. A sentence that wraps onto a second line is simply two unrelated runs of text that happen to sit one above the other. Nothing in the file says they belong together.

A Word document works the other way around. A DOCX file, defined by the Office Open XML standard (ECMA-376, later ISO/IEC 29500), stores paragraphs, styles, tables, section breaks and headers, and Word decides where each line breaks every time it lays out the page. When you add a word, the paragraph rewraps. Change the margin, the font or the paper size and every line moves. That behavior is exactly what you want when you edit, and exactly what a PDF never recorded, because a PDF was finished the moment it was saved.

So converting PDF to Word is reconstruction, not unpacking. The converter has to guess which lines form one paragraph, which bold line is a heading rather than a short bold sentence, which aligned words form a table, and in what order two columns should be read. Good converters guess well on clean documents and badly on busy ones. If the PDF was exported with tags (the structure layer used for accessibility), the guesses improve, because the tags record headings, lists and reading order explicitly.

There are two very different things that both end in .pdf. A text PDF was exported from Word, Google Docs, an accounting system or a browser's print dialog, and it contains real characters you can select, copy and search. A scanned PDF was made by a scanner, a copier or a phone app, and each page is a photograph of paper. To a converter, a scanned page holds no words at all, only pixels, even when the text looks perfectly crisp on screen. The file name and the icon tell you nothing; the two kinds look identical in a file manager.

A scanned page has to go through optical character recognition (OCR) before anything editable can come out of it. OCR looks at the shapes of letters and turns them into text, and its accuracy depends on resolution, contrast, skew and language. A clean 300 dpi black-on-white scan is read almost perfectly; a phone photo taken at an angle under a desk lamp produces typos in every paragraph. Converters without OCR either give you an empty document or a Word file with each page pasted in as one picture.

Many real-world files are mixed: a contract exported as text with a signed signature page scanned at the end, or a report with an appendix of scanned receipts. A converter that checks each page separately handles these best, because it reads the text pages directly and spends OCR only where it is needed, so the exported pages stay character-perfect. If you are making the scans yourself, the guide on scanning documents to PDF covers the resolution and lighting that give OCR its best chance.

Microsoft Word has opened PDFs directly since Word 2013, using a feature Microsoft calls PDF Reflow. It is the obvious choice if Word is already installed, and it does well on plain letters and reports. It converts one file at a time. Google Docs offers a similar route through Google Drive. Google's help page on converting PDFs to text recommends files of 2 MB or smaller and says lists, tables, columns, footnotes and endnotes are not likely to be detected, which matches what you see in practice: the words come through, the structure often does not.

Adobe Acrobat Standard and Pro export to Word through the Export PDF tool and run text recognition on scanned pages automatically. Acrobat is a paid subscription, and the free Acrobat Reader does not include the export. A browser converter sits between these: nothing to install, works on any operating system, and usually offers a choice between flowing text and exact layout. On the PDF to Word page, Text recognition (OCR) on Auto reads only the pages without a text layer, and Document language sets the OCR alphabet for scans that are not in English.

Word's PDF Reflow lives in the ordinary Open dialog, which is why many people never notice it. It works in Word 2013, 2016, 2019, 2021 and Microsoft 365 on Windows. Word shows a notice that it will convert the PDF into an editable document, that this may take a while, and that the result may not look exactly like the original, especially if the file has many graphics. Click OK and the converted copy opens as a new, unsaved document; the PDF itself is never changed.

Microsoft's own documentation lists elements that may not convert, including footnotes that span more than one page, endnotes, PDF bookmarks and comments, and font effects such as glow or shadow. In practice, Word also tends to turn sidebars, pull quotes and multi-column magazine pages into floating text boxes, which look right but are tiresome to edit. Word also has no setting for the language of a scan, so accented or non-Latin text on scanned pages is hit and miss. For a one-page letter or a plain report, though, it is often the quickest route there is, and the result already sits in the editor you will use.

Single-column body text with real headings converts well with any tool: letters, essays, contracts, most reports. Bold, italic, bullet lists and simple tables with visible borders usually survive. Images come through as long as the converter keeps them, which on the browser route is the Keep images option, on by default for DOCX. Hyperlinks generally survive when they were real links in the PDF; blue underlined text that was never a link stays plain text. The cleaner and more conventional the original layout, the less you will notice the conversion at all.

The trouble starts where the page layout carries meaning. Two-column text can be read across instead of down. Tables without ruling lines are hard to detect and may arrive as tab-separated lines. Headers, footers and page numbers are drawn on every page as ordinary text, so they often land in the middle of the body, once per page. Fillable form fields are flattened into plain text or lost. Embedded fonts that are not on your computer are swapped for the nearest match, which changes line lengths. For documents that are mostly tables, a spreadsheet is the better target; see extracting tables from images and PDFs.

Before you edit anything, spend two minutes checking the conversion, because a structural problem is easier to fix at the start than after you have rewritten half the text. Put the PDF and the Word file side by side and scroll through both. Look for missing paragraphs at page boundaries, headers repeated in the body and tables that turned into loose lines. With formatting marks on, a paragraph mark at the end of every line means you got a line-by-line layout instead of paragraphs; converting again with Flowing text usually fixes it faster than joining lines by hand.

For scanned pages, run a spelling check and pay extra attention to numbers, which spell checkers ignore. OCR commonly confuses the digit 0 with the letter O, 1 with l, and 5 with S, and in an invoice or contract a single wrong digit matters more than ten typos. If whole paragraphs are garbled, check the Document language and convert again before correcting by hand. When the layout is beyond repair, copy the clean text into a fresh document based on your own template; it is often faster than untangling text boxes.

Every converter is reverse-engineering a document that already existed in an editable form somewhere. If the PDF came from a colleague, a client or your own past self, the original DOCX, Google Doc or InDesign file is almost always better than any conversion: styles intact, tables real, tracked changes and comments preserved. One short email asking for the source can save an hour of cleanup, and people rarely mind. It also avoids a subtler problem: a converted file looks finished, so small errors introduced by the conversion can travel into the next version unnoticed.

Asking first is the right call for long documents you will edit heavily, for templates you will reuse, for anything with complex tables or forms, and for legal or financial documents where a single misread digit is a risk. Converting is the right call when the source is gone, when the PDF is a scan of paper that never had a digital original, or when you only need to lift a few paragraphs. In that case, the PDF converter also produces plain text, which is cleaner for quoting than a fully formatted Word file.

A signed or certified PDF is a special case. Converting it to Word removes the signature and its validity, so treat the Word file as a working copy only. If you need to send an edited version back, make your changes in the Word copy, export it to PDF again and get it signed again; do not present the converted file as the signed original. The same applies to PDFs with a certification or an official seal: the visual mark may survive, but its digital proof does not.