How to generate JATS XML from Word: options, common errors and how to validate it

Almost every scholarly article starts life as a Word file. But the databases that give a journal the most visibility — PubMed Central, SciELO, Redalyc — do not read Word: they require JATS XML, a format that describes the structure of the article in a way machines can understand. So how do you generate JATS XML from Word?

Getting from one to the other is one of the most common bottlenecks in editorial production. In this guide we review the options available, the most frequent errors and how to check that the XML is correct before submitting it.

What a JATS XML file needs to contain

JATS (Journal Article Tag Suite) is the ANSI/NISO Z39.96 standard. An article in JATS is divided into three main blocks:

• front: journal and article metadata (title, authors, affiliations, ORCID, abstract, keywords, dates, DOI, licence).
• body: the text, organised into sections, with its tables, figures, equations and citations.
• back: references, acknowledgements, funding, supplementary material.

The challenge is not just converting the text but correctly identifying every element: each author must be linked to their affiliation, each reference broken down into authors, title, journal, year and DOI, and each table must be a real table rather than an image.

Four ways to get from Word to JATS XML

1. Manual markup

Someone tags the article using an XML editor or the markup tools provided by some databases. It gives full control, but requires specific training and takes hours per article. It is also the method most prone to errors caused by fatigue.

2. External supplier

You send the final article and receive the XML. It is convenient, but there is a per-page cost and it adds lead time: every later correction has to be commissioned again, and the XML can fall out of sync with the PDF.

3. Generic plugins and converters

There are converters and plugins (for OJS, for example) that turn Word into XML. They work well with very clean documents, but tend to struggle with complex tables, unstructured references or inconsistent Word styles. The output almost always needs manual review.

4. Automated single-source production

Tools such as Index Production structure the article once from the Word file and generate the JATS XML, PDF, HTML and EPUB from that same source. The main advantage is that the formats never fall out of sync: a correction is applied once and reaches all of them.

Common errors that lead to rejections

A study published in 2025 in the journal Infonomy (Lázaro-Rodríguez) analysed JATS files from articles already published and found recurring errors. Among the most common:

• XML that does not validate against the JATS DTD or schema.
• Incomplete structure: expected elements are missing, such as affiliations, dates or the licence.
• Incorrect use of elements and attributes, or deprecated elements.
• Discrepancies between formats: the XML does not match the published PDF of the same article.
• Incorrect language declaration, especially in articles with abstracts in several languages.

Added to these are very common practical problems: references that are not broken down, unlinked figures, badly encoded special characters and tables converted into images.

How to validate the XML before submitting it

1. Validate against the DTD or schema for the JATS version required by the target database.
2. Use the official validators: PubMed Central offers file validation tools and a style checker, and the JATS4R initiative (NISO) provides a validator based on its best-practice recommendations.
3. Check the specific requirements of each database: SciELO, for example, works with its own JATS-based schema, SciELO PS.
4. Compare the XML with the PDF: title, authors, affiliations, number of tables and figures, and number of references should all match.
5. Review a rendered sample: viewing the XML converted to HTML helps you spot missing elements.

Which option suits your journal

If you publish only a few articles a year and someone on the team is proficient in markup, the manual method may be enough. If your output grows, or you also need to publish in HTML and EPUB, the biggest time saver is producing every format from a single source.

If you would like to see how one of your articles would look, you can request a trial with a manuscript from your journal on the JATS XML page. And if you want to know how much time you could save, try the savings calculator.