Convert a .md file to Word

What changes when you convert a file instead of pasting

The Markdown to Word converter on the home page is built around pasting. This page is built around a file you already have: a README in a repo, a note in an Obsidian vault, a Hugo or Jekyll post, a Pandoc source, an export from Notion or Bear. A .md file carries things a pasted string does not, and the converter treats them differently.

The file is read locally, with its name and encoding

When you drop a file, the browser hands it to the page as a File object. MDtoDOCX reads it with the FileReader API, so it never travels to a server. UTF-8 with or without a byte-order mark and UTF-16 are detected automatically; Windows CRLF line endings are normalised. The output takes the input's name, so weekly-report.md becomes weekly-report.docx, and the first heading or the front matter title becomes the document's Title property, which is what Windows Explorer and macOS Finder show in the details column.

Front matter becomes a title block and document properties

Most .md files that were written for a static site or a notes app begin with a YAML block between --- lines, or a TOML block between +++. Pasted into a converter that does not understand it, the first --- becomes a horizontal rule and title: Thesis Proposal becomes a stray paragraph at the top of the document. MDtoDOCX parses the block instead. title is written in Word's Title style and stored as the dc:title core property; author is stored as dc:creator and shown under the title; date is parsed, formatted for the document's locale and stored as the creation date. tags and keywords go into the Keywords property. Keys the converter does not recognise, such as draft, layout, permalink or weight, are never printed. Choose Properties only to keep the metadata but skip the visible title block, or Ignore to drop the front matter entirely.

Relative image paths

A file that says ![Baseline](./figures/baseline.png) refers to an image on your disk, and a web page cannot read your disk from a path. When you drop a single .md, the preview shows a placeholder with the path in place of each local image so you know what is missing. Drop the folder that contains the file instead, or select the images together with the .md, and the relative paths are resolved against what you dropped: the images are scaled to the page width and embedded in the .docx exactly as remote images are. Absolute https:// URLs are fetched as usual. Nothing you drop is uploaded; the folder is read in the browser.

Notes-app and static-site syntax

Files from Obsidian use [[wikilinks]], ![[embeds]], ==highlights== and > [!note] callouts. Wikilinks are written as their display text, embeds are resolved like relative images when the vault folder is dropped, highlights become Word's yellow highlight, and callouts become a shaded paragraph with the callout type as a bold label. Hugo shortcodes such as {{< figure >}} and Jekyll Liquid tags in {{ }} would print as literal braces in a document, so they are removed from the output; a comment listing them is added to the end of the document properties so nothing is lost silently.

Everything else is the same Word engine

Headings become Word's Heading 1 to Heading 6 styles, tables become real tables, fenced code keeps its indentation with xml:space="preserve", lists use Word numbering definitions, and footnotes become Word footnotes. The home page describes each of those in detail. Files up to 2 MB, roughly 300 pages, convert free in about a second. Pro adds an automatic table of contents, heading numbers, and batch conversion: drop a folder of .md files and receive a .zip with one .docx per file, names preserved, or one merged document in alphabetical order.

Frequently asked questions