Get the text of a PDF into an editable Word file

Turn the words of a PDF into a real .docx you can edit, free and with no signup. The text comes across; the layout does not, because a PDF never stored it.

Drop a file here

or paste from the clipboard

Options

On, the wrapped lines of a paragraph become one paragraph. Off, every line stays its own.

Starts each PDF page on a new page in Word. The page a line sat on is one of the few things a PDF really records.

Results appear here. Nothing is uploaded.

    How it works

    pdf.js walks the content stream and returns every text item with the transform that positions it, which the wrapper joins into lines. Those lines are then grouped into paragraphs using the only evidence available: blank lines, sentence-final punctuation followed by a capital, and hyphens at line ends, which are unwrapped because they are a typesetting artefact rather than part of the word. The paragraphs are written into an OPC package — a ZIP holding the content-type map, the root relationship and word/document.xml.

    What to watch out for

    The layout is not in the file to recover

    A PDF records glyphs with coordinates. It does not record paragraphs, columns, tables, headings or reading order, so none of those can be restored — they were never written down. Anything promising a faithfully reconstructed document is guessing, and guessing wrong on a table is worse than not trying.

    Tables arrive as lines of text

    A table in a PDF is text positioned in a grid, with no record that the grid exists. What comes out is the cell contents in the order the file lists them, run together as ordinary lines. For a two-column table that is often readable; for a wide financial table it is a mess, and rebuilding it by hand is the only route.

    A scanned PDF produces nothing, and says so

    A scan is a picture of a page, so there are no characters in it to extract. Rather than hand back an empty Word document that looks like the tool worked, the conversion stops and explains what the file is. Recognising those characters needs OCR, which this tool does not do.

    Paragraph joining is a guess, and can be turned off

    Deciding where one paragraph ends is inference, not fact. The rule used here under-merges deliberately, because two paragraphs wrongly joined take longer to fix than one wrongly split. If your document confuses it — poetry, code listings, forms — switch the option off and every line stays separate.

    Limits

    Files up to about 200 MB work reliably. Beyond that, the ceiling is your device's memory rather than any rule we impose — the work happens on your own machine. Tested in Chrome 140, Firefox 143, Safari 18.

    What you get, stated plainly before you start

    A .docx file that opens in Word, Google Docs, LibreOffice or Pages, containing every word of the PDF as editable paragraphs, with each PDF page starting a new page.

    What you do not get is the original document back. Not the fonts, not the columns, not the tables as tables, not the images, not the headings as headings.

    This is worth being blunt about, because most tools with this name are not. The reason is not effort or licensing — it is that the information is not in the file. A PDF is a description of marks on a page: put this glyph at this coordinate in this font at this size. Whether three of those glyphs are a heading, whether two columns are two columns or one wide one, whether a run of numbers is a table — none of that is recorded. A converter cannot recover it; it can only guess, and a confident wrong guess costs you more time than an honest plain result.

    So when is this the right tool?

    When you want the words. Quoting a report. Reusing a description you wrote last year. Translating a document. Marking up a contract. Getting a bibliography out of a paper. Handing text to someone who works in Word and does not want a PDF.

    For all of those, the layout was never the point, and a clean stream of editable paragraphs is exactly right.

    When it is the wrong tool: you need the document to look the same afterwards. Nothing that runs in a browser can do that, and the honest answer is to find the original file, or to rebuild the layout yourself, or to leave it as a PDF and add what you need to the PDF instead.

    How the lines become paragraphs

    A PDF has no paragraphs, so they have to be inferred, and the inference is visible rather than hidden.

    Text comes back as lines. A new paragraph is started when there is a blank line, or when a line ends in sentence-final punctuation and the next begins with a capital letter. A line ending in a hyphen is joined to the next without it, since that hyphen was a typesetter breaking a word across a line and not part of the spelling.

    The rule is deliberately cautious. It splits too often rather than too rarely, because merging two paragraphs that should be separate is much more annoying to undo than splitting one that should be joined. If your document defeats it — poetry, code, a form, a CV in columns — turn the option off and every line arrives as its own paragraph, which is a clean starting point for tidying by hand.

    Page breaks are real information

    Almost nothing about a PDF’s layout survives, but one thing genuinely does: which page a piece of text was on. That is recorded, not inferred.

    So each PDF page starts a new page in the Word file by default. It makes long documents navigable and gives you a landmark when comparing the two files side by side. Turn it off when you would rather have one continuous flow.

    Nothing is uploaded

    The PDF is read and the Word file is built in a Web Worker inside your browser. Contracts, invoices, medical letters, academic papers under review, internal reports — these are what people convert, and none of them need to be handed to somebody else’s server to have their words read out. The file never leaves your machine.

    Your files never leave your browser. Last updated 2026-08-02.