What this tool does
NexKit PDF Tools converts a PDF into an editable Word document by reading the text already embedded in the PDF and rebuilding it as real, editable paragraphs in a new .docx file - not a picture of the page, an actual Word document you can select, edit, and reformat.
This works on PDFs that already contain real text, such as PDFs exported from a word processor or generated from a document. It does not perform OCR: a scanned photo or image-only PDF has no text layer to extract, so this tool has nothing to convert for those pages - it detects that case and points you to OCR PDF instead of producing an empty or misleading result.
How to convert a PDF to Word
Open the PDF to Word page and choose a PDF. The browser reads the document, extracts text page by page, and then builds a .docx file, with a progress indicator for each of those stages so you can see what is happening and cancel before it finishes if you need to.
Open the downloaded .docx file in Word or a compatible editor and review the text before relying on it, especially for documents where exact layout, tables, or images matter - this tool converts text, not a page's visual design.
What PDF to Word can and cannot preserve
This tool preserves the words themselves and, in most cases, groups them into the same paragraph breaks as the source document, plus a page break between each original PDF page. That is the extent of what it reconstructs.
The PDF's overall page layout is not preserved either: rather than keeping each element at its original on-page position, the output reflows everything into a single top-to-bottom stream of paragraphs. Within that stream, it does not preserve fonts, font sizes, colors, bold or italic styling, or heading formatting - a large bold title in the PDF becomes an ordinary paragraph of plain text in the Word document, indistinguishable from body text once converted. Images are not carried over at all: any photo, chart, logo, or diagram in the source PDF is simply absent from the output, with no placeholder left behind. Tables are not reconstructed as tables; see the dedicated section below for what actually happens to them.
Why a PDF and a Word document are not the same kind of file
A PDF stores each page as a fixed arrangement of text and graphics at exact coordinates, designed to look identical everywhere it is opened. A Word document instead stores content as a flow of paragraphs, styles, and structure that a word processor lays out fresh every time, which is what makes it editable and reflowable in the first place.
Converting from one to the other means inferring structure - paragraphs, reading order, page breaks - from position data that was only ever meant to say where each character sits, not what role it plays in the document. This tool does that inference with a straightforward, disclosed heuristic based on line spacing and line width; it is not a full layout-analysis engine, so it will occasionally guess wrong on documents with unusual spacing.
Searchable PDF versus scanned PDF
A searchable PDF contains an actual text layer - character codes a computer can read - usually because it was exported directly from a word processor, publishing tool, or another program. This tool works on that kind of PDF.
A scanned PDF is fundamentally different: it is one or more photographs or scanned images of a page, saved inside a PDF wrapper, with no text layer at all - a computer sees only pixels, the same way it would see any other photo. This tool checks for extractable text before doing anything else, and if it finds none, it stops and directs you to OCR PDF, which is built specifically to recognize text in images and add a real text layer - after that, the result can be converted here.
Tables and multi-column layout limitations
Tables are not detected or rebuilt as Word tables. In practice, the cells of a table row typically collapse into one line of plain text with the original column spacing lost, and separate rows can merge into a single paragraph if their vertical spacing looks similar to ordinary paragraph line spacing to this tool's heuristic. The relationship between a row label and its values is easy to lose once that happens, so treat any table content in the output as a starting point to manually rebuild, not a working table.
Multi-column layouts (like a newsletter or a two-column report) are not detected either. The order text comes out in depends entirely on how the source PDF itself stores its content: some PDFs store column text grouped together (each column fully, in reading order), which happens to convert correctly, while others interleave text row by row across columns as it was visually laid out, which converts into a jumbled, out-of-order sentence combining fragments from both columns. There is no reliable way to predict which behavior a given PDF will have without opening the result and checking it.
Fonts, Chinese text, images, headers, footers, and page numbers
Chinese, English, and mixed Chinese/English text all extract correctly as text, character by character, regardless of which fonts the PDF used - the tool reads character codes, not the visual appearance of glyphs, so language and script are not a barrier by themselves. What is lost is the styling: font choice, size, color, and emphasis are not carried into the Word document, only plain text.
Images are always dropped, with nothing left in their place. Headers, footers, and page numbers are not recognized as special regions - they are extracted the same as any other text and appear as ordinary paragraphs mixed in with the page's body content, typically once near the top and once near the bottom of each page's converted text, rather than being removed, deduplicated, or placed in a real Word header/footer area. Page rotation itself is not a problem: rotated pages still extract their text in correct reading order, since the underlying text position data accounts for rotation on its own.
Encrypted and damaged PDF limitations
Password-protected or permission-restricted PDFs cannot be opened by this browser-based tool at all, so they are rejected immediately with a clear error rather than attempting a partial conversion - remove the password in the application that created the PDF, or with a dedicated unlock tool, then try again with the unlocked copy.
A damaged, truncated, or non-PDF file is rejected the same way, with a message that does not pretend to have converted something it could not read. This tool also has a 200-page limit and a file-size limit shown on this page; a PDF over either limit is rejected before any conversion work begins, rather than partially converting and cutting off partway through.
Troubleshooting
If you see the "no selectable text" error, the PDF has no text layer - this is not a bug, it means the pages are images (a scan or a photo). Follow the link to OCR PDF, run that tool first, and convert its output here.
If a table or multi-column page came out jumbled, that is the paragraph-inference heuristic doing its best with position data alone, not a crash - manually reformat that section in Word rather than expecting a fix by re-running the conversion, since the result depends on the source PDF's own internal structure, not on anything adjustable in this tool. If conversion fails on a large file, try splitting it into smaller PDFs first; if you cancel partway through, your original file selection is preserved so you can adjust settings or simply try again.
Privacy and file handling
Supported PDFs are processed locally in your browser, including text extraction and building the .docx file. Your PDF and the resulting Word document are not uploaded to NexKit servers for this workflow.
Keep your original PDF until you have confirmed the converted Word document meets your needs.
Frequently Asked Questions
Are my PDF files uploaded?
No. Supported PDFs are processed locally in your browser, including text extraction and building the Word document. The file is not uploaded to NexKit servers by this tool.
Does this work on scanned PDFs or photos of documents?
No. This tool extracts text that is already embedded in the PDF. Scanned or image-only PDFs have no text layer, so there is nothing to extract, and the tool detects this and points you to OCR PDF instead of producing an empty file. Run OCR PDF first to add a searchable text layer, then convert the result here.
Will the Word document look exactly like the PDF?
No. This tool reflows plain text into paragraphs and keeps page breaks between the original pages, but it does not preserve fonts, colors, images, or exact layout, and it does not rebuild tables as tables.
What happens to tables and multi-column layouts?
Table cells typically collapse into a single line of text with column structure lost, and rows can merge together. Multi-column text may come out in the correct reading order or may interleave into a jumbled sentence, depending on how the specific PDF stores its content - review any table or multi-column page before using it.
Will images and headers/footers be included?
Images are never included - they are dropped entirely with no placeholder. Headers, footers, and page numbers are not recognized as special regions; they are extracted like any other text and appear as ordinary paragraphs mixed into each page's content.
Does Chinese text convert correctly?
Yes. Chinese, English, and mixed Chinese/English text all extract correctly as text, character by character. What is not preserved is font styling - only plain text carries over, not the original typeface, size, or emphasis.
Does it support password-protected PDFs?
No. Password-protected and encrypted PDFs are rejected with a clear error, since this browser-based tool cannot open them at all.
How many pages can I convert at once?
This MVP supports PDFs with up to 200 pages and the file size limit shown on this page.
Can I cancel a conversion in progress?
Yes. A progress indicator shows the current stage, and cancelling stops the conversion and keeps your selected file in place so you can try again or choose a different file.