Extract the text from a PDF
Pull the words out of a PDF with the structure intact. Paragraphs rejoin instead of breaking mid-sentence, headings stay headings, and the running header and page numbers are left behind. Download as .txt or .md.
or tap to browse
Keeps headings, lists and paragraphs. Works on PDFs with a text layer.
Why most extracted text is unreadable
A PDF does not store paragraphs. It stores glyphs at coordinates, and a converter that simply prints them in order produces text that snaps at every line ending, repeats the document title on every page, and flattens every heading into an ordinary sentence. The structure is still recoverable, because the geometry carries it: headings are set larger, a wrapped line reaches the right margin while a finished one stops short, and a running header sits at the same height on each page. This tool reads those positions before it writes anything out.
The two kinds of PDF
Some PDFs contain real text you can already select in a reader. Others, particularly anything from a scanner or a photocopier, contain nothing but a picture of each page. This tool reads the first kind, which is the great majority, and does it instantly with no page limit. If your document is a scan, it will say so rather than hand back an empty file, and our Image and PDF to Text tool recognises those with OCR.
Text or Markdown
Plain text gives you the words with nothing added, which is what you want for pasting into a form or a search. Markdown keeps the shape: hash marks for headings, hyphens for bullets, blank lines between paragraphs. Choose Markdown if the result is going into notes, a wiki, a static site, or into a conversation with an AI model, where the structure is the difference between a document and a wall of text.
Frequently asked questions
Why is the text broken up mid-sentence in other converters?
Because they print each line of glyphs as its own line of text. A sentence in a PDF is laid out across several visual rows, and without checking whether a row reached the right margin there is no way to tell a line wrap from the end of a paragraph. This tool makes that check, which is why sentences come back whole.
Does it work on scanned PDFs?
No, and it says so instead of returning nothing. A scanned page is an image with no text in it at all. If you can select and copy words when you open the PDF in a reader, this tool will read it; if you cannot, use our Image and PDF to Text tool, which recognises the characters from the picture.
Is there a page limit?
No. A two-hundred-page report takes seconds, so there is no need to split a long document into parts first.
What happens to tables?
The cells come through as text in reading order, row by row. Rebuilding a grid reliably from coordinates is genuinely difficult, and a table that is silently wrong is worse than one that is plainly just text, so the words are kept and the structure is not invented.
Something is missing from the output.
Turn off "Skip headers and page numbers" and run it again. That setting leaves out lines that repeat in the same place in the top or bottom margin on most pages, which is right for a long report and occasionally wrong for a short document where a real line happens to repeat.