Convert TEX to DOCX: Move LaTeX Content Into Word

TEX-to-DOCX Is a Translation Between Two Editing Models, Not a Page Screenshot

A TEX file tells a TeX engine how to typeset a document; a DOCX file is a ZIP-based Office Open XML package containing paragraphs, runs, styles, relationships, and other Word structures. A useful conversion tries to recognize LaTeX sections, emphasis, lists, tables, citations, mathematics, and images, then writes editable Word equivalents. This is different from exporting a PDF. A PDF preserves a finished page image and geometry; DOCX should preserve enough structure that someone can revise paragraphs, add comments, and use Word's own styles afterward.

Pandoc describes its core design as readers that parse an input format into a native document representation and writers that turn that representation into a target format. That model explains both the strength and the limit of TEX-to-DOCX: common semantic constructs can be mapped well, while source commands that only make sense inside a particular LaTeX package may have no Word equivalent. Begin with the source project, not the compiled PDF, because a PDF conversion must first guess which visible glyphs were headings, list items, equations, or table cells. The original TEX offers those structural clues directly.

What Maps Cleanly Into Word and What Must Be Rebuilt After the Transfer

A simple LaTeX article usually has a strong migration path: \section and \subsection can become Word heading styles, ordinary paragraphs become editable paragraphs, \emph can become italic emphasis, and straightforward lists can become Word lists. A converter can also place local image files into the DOCX package when it can locate them. Pandoc's manual documents a --resource-path option for finding files that are not in the immediate working folder; an incorrect resource path is a real reason an output can contain missing figures even though the TEX source itself names them correctly.

The difficult material is package-specific or presentation-specific. LaTeX lets an author define commands, alter counters, place content in floats, draw with TikZ, create multirow table layouts, and invoke raw backend commands. Word has no general “run arbitrary LaTeX package macro” field. Pandoc's documentation is explicit that raw inline LaTeX is ignored in output formats other than Markdown, LaTeX, Emacs Org, and ConTeXt; raw TeX written for a typesetter should therefore be expected to vanish or need a substitute in DOCX. Treat the resulting document as an editable review copy until a person has checked every important object and citation.

Six Preparation Steps for a Word Copy That Collaborators Can Actually Review

• Choose the root document: convert the file with the document class and body, not an \input fragment that depends on another file.
• Gather every resource: retain figures, bibliography databases, custom commands, and included chapter files in their expected relative paths.
• Replace private macros where practical: expand a custom \MyTerm command into its text or map it before conversion if its definition carries meaning.
• Resolve references in the source first: a clean, compiled LaTeX version gives reviewers a reference for citations, numbering, and figure order.
• Use styles, not manual formatting: map headings to Word heading levels so navigation panes and accessible outlines work after handoff.
• Keep TEX as the master: Word edits should be reviewed and deliberately applied back to source; DOCX is not a lossless round trip to LaTeX.

Word Styles and Reference Documents Control Appearance After Structural Conversion

DOCX is not just paragraphs with a filename. It has named Word styles for headings, body text, captions, footnotes, tables, margins, headers, and more. Pandoc supports a --reference-doc option so the output can take its styles, document properties, margins, page size, header, and footer from a chosen DOCX. This is useful when a department or publisher requires a familiar Word template. It should be used carefully: Pandoc's maintainers advise that the reference document be a modified version of a DOCX produced by Pandoc, because arbitrary Word files can make assumptions that lead to a broken or surprising result.

A reference DOCX controls appearance, not the order or meaning of imported content. The Pandoc documentation and maintainer guidance distinguish those roles: its contents are ignored while its styles and properties influence newly generated output. Build one small test TEX file containing a title, three heading levels, a captioned image, a table, an equation, a footnote, and a citation. Convert it with the proposed reference document and open it in the actual version of Word used by collaborators. Only then apply the style setup to a long manuscript. This catches a missing style or malformed header before it affects hundreds of pages.

Use Word's Navigation Pane as a fast structural audit after conversion. It should show the expected hierarchy if the LaTeX headings reached real Heading styles rather than only larger bold text. Turn on formatting marks when paragraph gaps look wrong, because an imported manual line break and a true paragraph break behave differently when a reviewer edits text. Save an untouched first DOCX output before changing styles by hand. That baseline lets you tell whether a later visual difference came from conversion or from a reviewer’s Word-template adjustment. Keep it with the conversion log.

Why Equations, Labels, and Pictures Can Misbehave After a Successful DOCX Export

Math is a clear example of “editable” not always meaning “identical.” An older documented Pandoc issue shows inline and unnumbered display equations converting correctly while display equations carrying \label could leave raw LaTeX, lose their equation number, and show label text instead of a resolved cross-reference. The exact lesson is not that every current converter has that exact bug; it is that numbered equations and references are a distinct test case. Review them separately from ordinary inline math. If a target DOCX needs Word-editable mathematics, test a representative long equation, an aligned display, a numbered equation, and each cross-reference before relying on the workflow.

For missing pictures, check the source path and working directory first. A converter can only embed an image it can locate; configure the resource path or restore the original folder layout, then convert again. For a figure created with TikZ or another TeX drawing package, the source may not be an image file at all. Compile that figure to a stable PDF or PNG first, insert it intentionally, and keep the original drawing source alongside it. For labels and citations, avoid trying to preserve raw \ref or package-specific citation commands as plain Word text. Generate a reference PDF from the LaTeX source and compare every numbered target in the DOCX against it.

LaTeX Source Compared With an Editable Office Open XML Document

Questions Before You Give a LaTeX Manuscript to a Word-Based Reviewer

Can a reviewer use Track Changes in the DOCX?
Yes, that is one of DOCX's practical advantages. But merge the accepted wording back into the TEX master deliberately; converting the edited DOCX back to LaTeX is not a dependable way to recover all original macros and structure.

Why is a custom command missing from the Word file?
It may be raw LaTeX or a private macro with no DOCX equivalent. Expand or replace it before conversion, or recreate its meaning with Word text, style, image, or equation objects after conversion.

How do I make the DOCX match a house template?
Use a tested reference DOCX to supply styles, margins, headers, and footers. Build it from a converter-produced DOCX where possible, then test it on a representative source document before mass use.

Why are equations or cross-references wrong when paragraphs look fine?
Those features need their own mapping from LaTeX numbering and labels to Word objects. Compare numbered displays, labels, citations, and references with a PDF made from the original source.

Should I convert a PDF to DOCX instead?
Prefer TEX when it is available. TEX records sections and source relationships; a PDF exposes final positioned content and forces the converter to infer the document structure again.