How it works
- This is the one tool on OneTapTools that needs a server, and the page says so before the file picker opens. Nothing about that is hidden behind a tooltip.
- You add a file. It is uploaded over TLS to a converter running pdf2docx, which reads the PDF with PyMuPDF and writes the .docx directly. It is a purpose-built reconstructor rather than an office suite driven from a command line.
- The conversion runs in an isolated process with no persistent storage attached. The converter reads every text span with its font, size and position, groups spans into lines, lines into paragraphs, and detects tables from both ruling lines and column alignment. Images are extracted and placed rather than dropped.
- What comes back is a Word document made of real paragraphs and real tables — not a page of positioned text frames, which is what you get from converters that route a PDF through a drawing engine. Cell borders, headings and list numbering survive as document structure you can edit.
- The .docx is streamed back as the response body and the uploaded source is deleted when the process exits, within seconds. Nothing is written to a database and no copy is kept — there is no history for you to clear because there is no history.
Limitations
Complex layouts convert imperfectly, and it is worth understanding why so the output does not surprise you. A PDF does not contain paragraphs. It contains instructions to draw glyphs at coordinates. Everything a converter gives you — paragraph breaks, reading order, table cells, list numbering — is inferred from where the marks landed on the page. Single-column prose infers cleanly. Multi-column pages, sidebars and pull quotes often interleave, because the reading order the converter guesses is not the one your eye uses.
Tables are where converters differ most. A table drawn with visible ruling lines reconstructs here as a real Word table with real cells. One laid out purely with whitespace alignment is detected from column positions and usually survives too, but it is the case most likely to come back as ordinary paragraphs — check any borderless table before you rely on it.
Scanned PDFs are images, and images contain no text to convert. If the PDF has no OCR text layer, the .docx you get back will be a picture of a document embedded in a Word file. Run OCR first — most phone scanner apps and Acrobat both do it — then convert.
Fonts substitute. A PDF embeds a subset of a typeface; a .docx references a font by name and expects the reader's machine to have it. Where the original is not installed, Word picks the closest match and line breaks move as a result — the words are right, the pagination may not be.
Questions
Is my file uploaded?
Yes, and this is the only tool here where that is true. Reconstructing paragraphs and tables from positioned glyphs is not something a browser tab can do. The file goes over TLS to an isolated process with no persistent storage, is converted, and is deleted when that process exits. If that is not acceptable for the document in question, do not use this tool.
Is there a file size limit?
Up to 40 MB, because the conversion happens on a server with a request timeout. A file larger than that will usually be a scan, and a scan is the case this tool handles worst anyway.
How long is the file kept?
It is not kept. It exists in the converter process for the duration of the conversion and is deleted when that process exits — seconds, not hours. There is no storage bucket, no backup and no queue it sits in.
Why did my tables come out wrong?
Because the table was drawn with alignment rather than ruled lines, so there was nothing in the file for the converter to recognise as a table. Tables with visible borders convert far more reliably. If the layout matters more than the text, it is often faster to retype the table than to repair it.
Do I need an account?
No. There is no sign-up, no email step and no usage counter. The tool is paid for by the ads on this page.
When people use it
The usual reason is that someone sent a final document and then asked for changes. A contract arrives as a PDF, three clauses need editing, and the person who has the original .docx is on leave. The conversion does not have to be perfect for that — it has to get the text into an editable shape so the changes can be made and the document re-exported.
The second reason is reuse. A report from last year contains four paragraphs and a table that belong in this year's version. Converting the whole document and taking the parts you need is usually faster than retyping, even accounting for the cleanup, provided the pages are single-column prose.
The usual next step after converting is checking what changed against the version you started from. Text diff compares the converted wording against the original line by line, without either version leaving your browser.
How it compares
This is the only tool here that needs a server, and the page says so before you add a file. If that is not acceptable, the merge, split and compress tools work entirely in your browser and never see the network. Against a desktop route: Word's own PDF import converts without an upload and is worth using if you have it — comparable on prose, weaker on tables. LibreOffice routes a PDF through its drawing engine and tends to produce positioned text frames rather than editable paragraphs, which is the specific outcome this converter was chosen to avoid.