Skip to content

TMX (.tmx) Translation Memory Files

Format and workflow for the OASIS Translation Memory eXchange standard.

TMX (Translation Memory eXchange) is an XML-based standard for exchanging translation memory between CAT tools and translation management systems. A .tmx file contains translation units — pairs of source and target text in one or more languages — plus metadata about who, when, and which tool produced them. While XLIFF moves work-in-progress translations between tools, TMX moves accumulated translation memory: completed translations stored for fuzzy-match reuse on future projects. TMX is the format you reach for when migrating between TMS platforms or backing up your translation memory.

Key facts
  • What it is: XML-based translation memory exchange format
  • Standard body: OASIS (originally LISA)
  • Most common version: TMX 1.4 (2005)
  • File extension: .tmx
  • Encoding: UTF-8
  • Used for: TMS migration, translation memory backup, sharing TM with translators

What a TMX file looks like

A TMX file declares its metadata in a <header> and lists translation units inside a <body>. Each <tu> (translation unit) has multiple <tuv> elements — one per language — each containing a <seg> with the actual text:

<?xml version="1.0" encoding="UTF-8"?>
<tmx version="1.4">
  <header creationtool="Locize" creationtoolversion="1.0" segtype="sentence"
          adminlang="en" srclang="en" datatype="plaintext" o-tmf="locize"/>
  <body>
    <tu tuid="welcome.title">
      <tuv xml:lang="en">
        <seg>Welcome, {name}!</seg>
      </tuv>
      <tuv xml:lang="de">
        <seg>Willkommen, {name}!</seg>
      </tuv>
    </tu>
    <tu tuid="login.submit">
      <tuv xml:lang="en">
        <seg>Sign in</seg>
      </tuv>
      <tuv xml:lang="de">
        <seg>Anmelden</seg>
      </tuv>
    </tu>
  </body>
</tmx>

Common TMX workflows

  • Migrating between TMS platforms. Export your translation memory as TMX from the old platform, import into the new one. Years of accumulated translations and fuzzy-match data move with you instead of being abandoned.
  • Backing up translation memory. Periodic TMX exports protect your TM as a separate, tool-independent backup.
  • Sharing TM with freelance translators. Export a TMX scoped to relevant content; the translator imports it into their CAT tool to get fuzzy-match suggestions while they work.
  • Bootstrapping a new project. Import an existing TMX into a fresh TMS project to seed the translation memory before any new translations are added.

How to use TMX in Locize

Locize imports and exports TMX files. Once imported, translation memory matches surface automatically in the CAT view as you translate — both fuzzy and exact matches:

  1. Get started
  2. Import your .tmx file
  3. The translation memory is now available across all your projects
  4. Export the TM back to .tmx for migration or backup

Frequently asked questions

What is a TMX file?

TMX (Translation Memory eXchange) is an XML-based standard for moving translation memory between CAT tools and translation management systems. A .tmx file contains translation units — pairs of source and target text, optionally in multiple languages — along with metadata like creation date, project, and tool. TMX is to translation memory what XLIFF is to active translation work.

When do I use TMX vs XLIFF?

XLIFF moves work-in-progress translations between tools (with status, comments, segment IDs). TMX moves accumulated translation memory — completed translations stored for fuzzy-match reuse on future projects. You typically export TMX once when migrating between TMS platforms, or periodically to back up your translation memory.

How do I import or export TMX?

Most CAT tools and TMS platforms have explicit import / export buttons for TMX. In Locize, .tmx files import and export through the same file-format pipeline as XLIFF and other formats. Translation memory matches surface automatically in the CAT view as you translate.

What is translation memory?

Translation memory (TM) is a database of previously translated source / target pairs. When you translate new content, the TMS searches the TM for matches — exact or fuzzy — and surfaces them as suggestions. This both speeds up translation and improves consistency: the same source text gets the same translation every time.

TMX versions?

TMX 1.4 (released 2005) is the most widely supported version. Earlier versions exist (1.1, 1.2, 1.3) but virtually all modern tools target 1.4. Unlike XLIFF, TMX has not had a major version revision recently — the format is stable.

Bring your translation memory with you.
Import TMX, build TM as you translate, export TMX whenever you need to migrate or back up.