PDF to Word Converter

Choose a PDF, or drop it on the page. It opens straight away because nothing is uploaded. Up to 50 MB and 300 pages.

Decide how the Word file should be built. Leave it on Decide for me and the tool works it out from the page itself, or choose Flowing text for a report or a letter you will edit heavily, or Match the PDF layout for a form, invoice, receipt or certificate where the layout is the point.

Look at the outline on the left. It shows the headings, paragraphs, lists and tables the tool found. If any pages are scans it says so, and those will be read with OCR when you convert.

If anything looks wrong, open the panel on the right that covers it - Text, Pictures and look, Tables, Scanned pages or Pages. They start closed and only one opens at a time, so the page stays short. Most documents need nothing changed here.

Press Convert to Word. Scanned pages are read first, with a progress bar, then the .docx downloads and opens in Word, LibreOffice, Google Docs or Pages.

No. The reading and the conversion both happen inside your browser, on your own device, so the file never leaves your computer. There is no upload wait, no queue, nothing stored on a server and nothing for us to delete. That matters more for PDFs than for most files, because the documents people convert tend to be contracts, invoices, statements, medical letters and CVs.

Yes. Tables that have drawn borders are read from the border lines themselves, which is accurate rather than guesswork, and come out as real Word tables you can edit. Logos and photographs are extracted and placed at the right size. Charts, diagrams, boxes and arrows drawn as vector graphics are captured as pictures and put back in position - they look right but cannot be edited as shapes, because Word has no equivalent that a converter could produce sensibly. Text colour, the original font sizes, centred and right-aligned lines, and the page size and orientation all carry across too.

That depends on which of the two modes you use, and it is the most important choice on the page. Match the PDF layout is the one that looks like the original: it draws the coloured panels, bars and rules again as shapes and places each line of text at the position it really had, so an invoice or a receipt comes out looking like the invoice or receipt you started with - and the text is still real text you can click into and change. Flowing text does not try to look identical, on purpose. It rebuilds the document as paragraphs, headings, lists and real Word tables, so it reflows properly when you type, which is what you want in a report or a letter. Neither mode is a photograph of the page: fonts are substituted, so lines are measured and gently squeezed or stretched to end up the width they had, and very unusual page furniture can still differ. Leave the choice on Decide for me and the tool reads how form-like the page is and picks - and it tells you which one it chose.

Choose Match the PDF layout and convert it again. A form is not a flowing document: it is text placed at coordinates over some coloured rectangles, and the structure it appears to have was never really in the file. Rebuilding one as paragraphs and tables is what splits words across cells and drops the coloured section bars. The layout mode stops guessing and redraws the page instead, which is why forms come out right in it. Decide for me normally picks it on your behalf, but if a document sits on the line you can force it.

Yes, and that is the whole point of it. Each line sits in its own frame, so you click into it and type exactly as you would anywhere else in Word - change a name, correct an amount, fix a date. What you do not get is reflow: because every line is positioned, typing a long sentence into one frame will not push the lines below it down the page the way it would in an ordinary document. For heavy rewriting, convert again with Flowing text.

Headings are worked out from size: the tool finds the size most of the body text uses, then treats noticeably larger lines as headings, with bigger jumps becoming higher levels. Bold and italic come from the actual font each piece of text is set in, read from inside the PDF, rather than from guessing. If either gets it wrong on your document you can switch that detection off and convert the plain text instead.

Yes, and they come out as real Word tables you can edit, not as loose text. The tool finds rows by looking for text that lines up in the same columns on consecutive lines. That works well on ordinary tables with clear column gaps. It is less reliable with merged cells, nested tables, or tables whose columns are only implied by ruled lines with text that does not align. If a table comes out wrong, turning table detection off gives you the text as ordinary paragraphs, which is sometimes easier to fix by hand.

Yes. If a PDF has no text in it - because someone scanned, photocopied or photographed the page - the tool reads the letters out of the image instead, using OCR. It spots which pages need this and tells you before you convert, so a scan is never quietly turned into a blank file. Reading a page takes a second or two, and the recognition engine downloads once the first time you use it and is then cached by your browser. Everything still happens on your own device; the scan is not uploaded.

On clear typed documents it is very good. In testing on a scanned page, including a deliberately degraded version that was skewed, blurred, washed out and heavily JPEG-compressed, every word was recovered correctly. Where it struggles is handwriting, which it cannot read reliably at all, very faint or low-resolution scans, unusual or decorative fonts, and heavy background patterns. Always read the result through - OCR mistakes tend to be plausible-looking words rather than obvious nonsense, so they are easy to skim past.

Twelve: English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Turkish, Indonesian, Arabic and Urdu. Pick the one your document is written in before converting, because the recognition model is different for each. A document in the wrong language still produces text, but a lot of it will be wrong.

Both. Automatic is the default and uses OCR only on pages with no text in them. Never skips it entirely, which is useful when you only want the pages that already contain text and do not want to wait. Every page forces it on, which helps with the awkward middle case: a PDF that has a thin layer of bad text over a scan, where the existing text is worse than what OCR would produce.

Because the tool has to decide which line breaks in the PDF were real paragraph endings and which were just the text wrapping. It joins a line to the one above when the spacing is normal, the previous line ran the full width of the page, and both start at the same margin. Documents with unusual layouts can fool that. If it joins too much, untick "join lines into paragraphs" to keep every line separate; if it splits too much, the same setting is the one to experiment with.

Anything that reads .docx: Microsoft Word 2007 and later, LibreOffice Writer, Google Docs, Apple Pages, WPS Office and most phone office apps. It is a standard Word file, not a renamed PDF or an RTF pretending to be one.

No watermark, no account, no email address and no cap on how many files you convert. The only limits are practical ones: 50 MB and 300 pages per file, because everything is processed in your browser using your own memory.