Extract the text from a PDF

Pull the words out of a PDF as plain text, free and instantly. Honest about why this is not a Word document and what a scanned file cannot give you.

Drop a file here

or paste from the clipboard

Options

Useful for lifting one chapter out of a long document.

Turn this off when the text is going somewhere that should read continuously.

Results appear here. Nothing is uploaded.

    How it works

    pdf.js returns the text of a page as a list of runs, each with a transform matrix saying where it sits. There are no lines in that data, because a PDF never recorded any — so runs are grouped by the vertical position in their matrix, rounded to the nearest point, and each group is sorted left to right before being joined. That reconstruction is a heuristic, and it is the reason a two-column layout can come out interleaved.

    What to watch out for

    A scanned document has no text to give

    If the PDF is a photograph of paper, the characters exist only as pixels and nothing here can read them. This tool detects that case and says so rather than handing back an empty file. Turning those pixels into characters is optical recognition, a different and much heavier job.

    Columns, tables and footnotes come out in reading order only by luck

    Lines are rebuilt from vertical positions, so a two-column page interleaves left and right, and a table becomes a run of cell contents with no structure. Single-column prose extracts cleanly; anything laid out in blocks needs a human pass afterwards.

    Formatting is gone, not hidden

    Bold, italic, headings, lists and font sizes are properties of how glyphs were drawn, and none of them survive into plain text. What you get is the words in order. That is the honest ceiling of this operation, not a limitation of this particular tool.

    Limits

    Files up to about 100 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.

    Why this is not “PDF to Word”

    It is the closest thing that can honestly be built, and the gap is worth explaining, because a great many tools promise otherwise.

    A PDF does not contain paragraphs. It contains instructions of the form “draw these glyphs at this position in this font”. Whether two runs of glyphs are the same sentence, whether a block is a heading, whether a grid of numbers is a table — none of that is recorded anywhere in the file. A word processor needs all of it.

    Reconstructing that structure is a layout-recognition problem, and it is one that even expensive commercial software solves imperfectly. What can be done reliably is to recover the characters, which is what happens here.

    If the goal is to quote a passage, search a document, feed text into something else, or count words in a submission, this is the right tool. If the goal is a document you can restyle, no browser tool will get you there and the ones that claim to are producing a box of loose text with the formatting invented.

    Why lines come back the way they do

    The text of a page arrives as runs with coordinates. Lines are rebuilt by grouping runs that share a vertical position and sorting each group from left to right.

    For ordinary single-column prose that reconstruction is almost always right. For a page in two columns it is almost always wrong, because a line on the left and a line on the right sit at the same height and get joined. Footnotes, marginal notes, headers and page furniture arrive wherever their coordinates put them.

    Knowing that in advance makes the output far more useful: extract a chapter of running text and it is clean, extract a page of a journal in two columns and expect to do some tidying.

    When there is nothing to extract

    A large share of the PDFs people have are scans — a photograph of a page wrapped in a PDF container. There are no characters in such a file at all, only pixels arranged to look like characters.

    This tool checks and tells you, rather than returning an empty file and leaving you to work out why. Getting text out of a scan requires optical character recognition, which is a much heavier piece of machinery and a genuinely different tool.

    A quick way to tell before you start: open the PDF in any reader and try to select a sentence with the mouse. If nothing highlights, it is a scan.

    Nothing is uploaded

    The document is parsed in a Web Worker in your browser, and the text never leaves it. Contracts, theses, medical letters and legal filings are exactly the documents people extract quotes from, and exactly the ones that should not be posted to a server to do it.

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