How to Convert a PDF to Word — and What Actually Survives

Convert a PDF to Word in four steps

1. 1
   Open the converter. Go to pdfgami.com/pdf-to-docx. There is nothing to install, no account to make, and no watermark waiting at the end. Before the dropzone you will see a panel saying the tool carries the words and not the layout — that notice is there on purpose, so the limits arrive before you spend time rather than after.
2. 2
   Drop the PDF. The file is read in your own tab and the page count appears next to the filename. If the PDF is encrypted, damaged, or not really a PDF, you get a plain message saying so instead of a silent failure.
3. 3
   Choose pages and settings. Leave Pages empty for the whole document, or type something like 1-5, 9, 12-14. Three checkboxes control the shape of the output: reflow paragraphs, a page break between PDF pages, and a small label above each page. The defaults are the right answer for most documents.
4. 4
   Convert and download. Click Convert & download .docx. A progress ring tracks the pages as they are parsed, the Word file is assembled in the tab, and it lands in your downloads folder named after the PDF. Afterwards the tool reports how many words came across from how many pages — and warns you separately if any of those pages turned out to be scans.

What survives the conversion, and what does not

This is the section most guides bury and it is the only one that decides whether the tool is useful to you. Here it is flatly.

Reading order deserves a note, because it is the difference between output you can use and output you throw away. pdf.js — the PDF engine this tool loads into your tab — hands back a flat list of positioned text runs in whatever sequence the PDF generator happened to write them. On a two-column academic page that sequence often alternates between the columns, which is exactly why so much “PDF to text” output reads like interleaved nonsense. The extractor here rebuilds the order from geometry instead: it looks for bands of whitespace that run clear across a block and cuts there, recursively, so a full-width title is peeled off first and the two columns underneath are then found and read one after the other. Lines are grouped by shared baseline, word spaces are inferred from horizontal gaps, and a vertical gap noticeably larger than the normal leading is taken as a paragraph break.

So the columns are understood — that is what makes the prose come out in the right order — and then deliberately flattened into one running sequence of paragraphs. Your two-column paper becomes a single-column Word document with the sentences intact. That is the honest maximum: the structure was used to recover the text, not to reproduce the design.

Why layout cannot be rebuilt honestly

It helps to know what a PDF actually is. A Word document describes intent: this is a heading, this is a table with four columns, this paragraph belongs to a bulleted list. A PDF describes result: place this glyph at these coordinates in this font at this size, then the next one. It is a finished layout, closer to a printing plate than to a document. That is precisely why PDFs look the same everywhere — and precisely why going backwards is guesswork.

Converting a PDF to Word properly means inventing the intent again from the result. Were those three runs of larger text a heading, or an emphasised sentence? Do those numbers sitting in vertical alignment form a table, or are they a list with a tab stop? Where does a cell begin when there are no ruling lines to read? Is that bold word a style, or a different embedded font that happens to be heavier? Every one of those questions has to be answered by heuristic, and each heuristic is wrong some of the time.

There is a second, more practical reason. Everything on PDFGami runs as JavaScript in your own browser tab, and the libraries that do credible layout reconstruction are large, commercial, server-side, and licensed accordingly. A browser-only tool that claimed full fidelity would be claiming something the shipped code cannot do. So this one does the part it can do well — recover the text faithfully and write a clean .docx — and says out loud what it leaves on the table.

Reflow, page ranges and page breaks

Reflow paragraphs is the setting that matters most, and it is on by default. A PDF line ends where the original page was wide enough to need a break — not where your Word page needs one. Carry those breaks through literally and you get a document where every line is its own stub paragraph, ragged down the page, impossible to edit without deleting breaks by hand. With reflow on, the lines of a paragraph are joined into one paragraph and Word rewraps them to its own page width, the way text is supposed to behave. Turn it off only when the line breaks themselves are the content — poetry, code listings, addresses.

Page break between PDF pages is also on by default, and starts each PDF page on a fresh Word page. Useful when the page boundaries are meaningful, such as a form or a report you will keep comparing against the original. Switch it off when you want one continuous flow of text and do not care where the PDF happened to break. Label each page is off by default; turn it on and a small grey Page 7line appears above each page's text, which is handy when you are going to cite page numbers.

The Pages field takes an empty value for everything, or the standard shorthand: 1-5, 3, 1,3,5-7. It is checked while you type: a page outside the document, or a range the wrong way round, is reported straight away, and otherwise the field tells you how many pages are selected. Duplicates are collapsed and the pages come out in order regardless of how you typed them.

Scanned PDFs have no text to convert

If your PDF came out of a scanner, a photocopier or a phone camera, it is very likely a stack of page images with no text layer at all. You can see words; the file does not contain any. The quick test takes five seconds: open the PDF and try to select a sentence with your mouse. If the selection highlights text, there is a text layer and the conversion will work. If you get a rectangle over a picture, there is nothing to extract.

The tool does not pretend otherwise. For every page it checks whether meaningful characters came back, and if not, whether the page paints any images — which separates “this is a scan” from “this page is genuinely blank”. Each such page gets a bracketed note written into the Word file saying the page is a scanned image and holds no text, rather than arriving as a mysterious empty gap, and the summary afterwards tells you how many pages were affected.

Getting words off a picture requires OCR — optical character recognition — and PDFGami does not do OCR. For a scanned document you will need a tool built for that job, and you should expect to proofread its output, since OCR misreads characters in ways that spellcheck will not always catch.

When you genuinely need the formatting back

Sometimes text is not enough — you need the branded template, the real table, the invoice that still looks like an invoice. In that case here are the routes that actually work, with their costs stated rather than glossed.

• 1.Find the source file. Every PDF was something else first — a Word document, a spreadsheet, an InDesign layout. That original still has the structure the PDF threw away, so recovering it beats reconstructing it every time. Ask whoever sent the PDF before you do anything clever; this works more often than people expect, and it is the only option that costs nothing and loses nothing.
• 2.Adobe Acrobat's paid export. Acrobat's PDF-to-Word export is generally the strongest of the reconstruction tools, since Adobe wrote the format. It is a paid subscription, and where the conversion actually happens differs between the desktop app and Adobe's web tools — worth checking before you hand it a confidential document. It also still guesses: complicated tables and multi-column layouts come back needing cleanup. It is the honest recommendation when fidelity genuinely matters, not a magic answer.
• 3.Retype a short document. Unglamorous and frequently the fastest path for one or two pages, especially a form or a page with a complicated table. Convert with the tool here to get the words, then paste them into your own template and lay them out. Fifteen minutes of typing beats an hour of fighting a mangled reconstruction.

And one route that is often overlooked: if you only needed to read or quote the content, you may not want Word at all. The PDF to text converter gives you the same carefully ordered text as a plain .txt file, which is simpler to search, diff or feed to something else. If the PDF is a deck rather than a document, PDF to PowerPoint is the better destination.

Free, and nothing is uploaded

“Free PDF to Word” usually means free in exchange for your document. The standard architecture is an upload form, a converter running on someone's server, and a download link — which puts a copy of your file on a disk you do not control, governed by a retention policy you did not read. For a takeaway menu that is fine. For a signed contract, a payslip, a medical letter or a CV with your home address in it, that is a bad trade for saving a minute.

Here the whole conversion is JavaScript running in your tab. Your browser parses the PDF, the text is recovered locally, and the .docx is assembled locally too — a .docx being a ZIP of OOXML parts, which the page builds with JSZip: the content-type map, the package relationships, a small styles part and the document body. It is deliberately the minimum that Word, LibreOffice and Google Docs all accept, which is why the file opens cleanly in all three without a repair prompt. The download comes from a blob created in the page, not from a server.

Two consequences worth knowing. There is no queue and no file size limit imposed by anyone else's hardware — the practical ceiling is your own device's memory, since the PDF and the document being built both live in the tab, so a 600-page file will go further on a laptop than on a phone. And no request ever carries the document itself, which is the one part of a no-upload claim you can check for yourself rather than take on trust.

Do not take that on trust from us any more than from anyone else. Open your browser developer tools, switch to the Network tab, and run a conversion: an uploading tool shows a request the size of your file, and this one stays quiet. If you want the longer version of how to evaluate any file tool before you feed it something confidential, we wrote it up in are online PDF tools safe.

Frequently asked questions

Continue reading