# How to Convert a TMX to Excel and Back

Source: https://traduq.com/blog/convert-tmx-to-excel · Updated November 18, 2026

## Key takeaways

- A spreadsheet holds the text of a TMX well. It doesn't hold properties, notes, dates or users, so the TMX you build from it back is a new file, not the old one with edits.
- Inline tags survive the trip only if nobody edits the tag code in the cells. Edit the words around them and leave the XML alone.
- Save the reviewed file as .xlsx, not CSV. Reopening a CSV in Excel is where leading zeros, long numbers and some dates get converted.
- Excel cells stop at 32,767 characters and a sheet at 1,048,576 rows, per Microsoft. Most memories fit, but long segments or huge memories may not.
- To fix a few units and keep every property, edit the TMX itself instead. Use the spreadsheet for review, sharing and bulk work on the text.

Turn a TMX translation memory into an Excel sheet to review or share it, then convert the edited sheet back to TMX without losing tags or languages.

To convert a TMX translation memory to Excel, open it in a converter that puts each language in its own column, review or edit the sheet, and convert the saved .xlsx back to TMX with the same language codes in the header row. The free [TMX to Excel converter](https://traduq.com/tools/tmx-converter) does both directions in your browser. The trip is safe for the text and the tags. It is not safe for the metadata, so read the part about what gets lost before you replace anything.

I get asked for this more than I expected. A client wants their in-house reviewer to check a memory, and the reviewer doesn't have a CAT tool. A PM wants to see what's in a TM before buying more work on top of it. I want to sort 20,000 units by target and look for the same mistranslation repeated 40 times. Everyone has Excel, or at least something that opens .xlsx. Nobody wants to read raw XML, and I don't blame them.

I checked every external page linked here on 17 November 2026 (see Sources at the end).

## What a spreadsheet keeps and what it can't

A TMX is more than pairs of sentences. The [TMX 1.4b specification](https://www.gala-global.org/tmx-14b) splits it into structural elements (units, language variants, properties, notes) and inline elements, the codes for bold, links and placeholders inside a segment. Each unit can also carry who created it and when, who changed it last and when, and properties like client or project.

A sheet with one column per language holds the text. Everything else has nowhere to go. In practice, this is what happens.

- **Text** comes through as it is, one row per unit and one column per language.
- **Inline tags** come through as text, as their XML. They're ugly, but they survive if nobody edits them.
- **Unit IDs** can survive as an extra column, if the TMX has them.
- **Properties, notes, dates and users** don't survive. When you convert the sheet back, you get a clean new TMX with today's header and no metadata.

That last point decides how you use the result. A reviewed sheet is perfect for checking and correcting translations. It isn't a way to edit a memory in place.

## How to convert a TMX to Excel

1. Open the [TMX to Excel converter](https://traduq.com/tools/tmx-converter) and drop your TMX. TMX files from Trados, memoQ, Phrase, Wordfast, OmegaT and others work, up to 200 MB.
2. Check the preview. Every language in the file becomes a column, with the header's source language first. A multilingual TMX gives you one column per language, not one file per pair.
3. Under "Convert to", choose Excel (.xlsx). Pick the source column and the target columns you want.
4. Choose what to do with inline tags. "Keep" leaves them in the cells as XML so they come back later. "Remove" gives clean text, which is nicer to read but can't go back to TMX with its formatting.
5. Optionally tick "Skip segments with no translation", then click "Convert and download".

The sheet opens with a frozen header row, wrapped text and wide columns, and every cell is written as text, so Excel doesn't turn "1-2" into a date when it opens the file. If any segment is longer than Excel allows, the converter tells you how many cells it had to cut.

## Editing the sheet without breaking it

This is where most round trips fail, and it's never the software's fault. It's the reviewer who helpfully "cleaned up" the weird code.

- **Leave the tag code alone.** A cell may look like **Click <bpt i="1">&lt;b&gt;</bpt>Save<ept i="1">&lt;/b&gt;</ept>**. Edit "Click" and "Save", not the brackets. If a tag is damaged, the converter can't recognise it, and it comes back as plain text inside the segment instead of a tag.
- **Keep the header row.** The language codes in row 1 (en-US, es-MX and so on) are how the converter knows which column is which. Codes or plain names like "English" and "Spanish" both work.
- **Keep the ID column** if there is one. That's how each unit keeps its ID on the way back.
- **Sorting and filtering are fine.** Each row is a whole unit, so the order doesn't matter.
- **Don't add comment columns in the middle.** To the converter, a reviewer's notes column looks like another language. If you need one, put it at the end and leave it out of the target columns when you convert back.
- **Line breaks inside a cell** are Alt+Enter in Excel for Windows ([Microsoft](https://support.microsoft.com/en-us/office/start-a-new-line-of-text-inside-a-cell-in-excel-33e41eab-8b5e-4193-93d6-9a06ecf812b3)). Pasting text with line breaks straight onto the sheet can spread one segment over several rows, so paste it inside the cell (F2 first).
- **Only the first sheet is read.** Put your notes on a second sheet if you like.

## How to convert the Excel file back to TMX

1. Save the file as .xlsx. Not CSV (more on that below).
2. Drop the .xlsx in the [same converter](https://traduq.com/tools/tmx-converter). It reads the first sheet and detects the header row.
3. Check that every column has a language. If a header isn't a language the converter knows, type the code (en, es-ES) before converting.
4. Choose TMX, pick the source and target columns, and click "Convert and download".

You get a TMX 1.4 file in UTF-8 with one unit per row. Empty cells are left out, so an empty translation doesn't become an empty unit.

Then decide what to do with it. I import it into the CAT tool memory as an update and keep the original TMX as a backup. How the import treats units with the same source (overwrite, add, keep both) depends on the tool's import settings, so check them before a big import. What I don't do is throw away the original memory and replace it with the round-tripped file, because the dates, users and client properties stay behind in the original.

## The Excel traps

None of these are bugs. Excel is a calculator that's being polite about your text.

**Automatic conversion.** Microsoft says Excel is "preprogrammed to make it easier to enter dates", so 12/2 becomes 2-Dec, and that this can't be turned off. Its fix is formatting the cells as Text or starting the entry with an apostrophe ([Microsoft](https://support.microsoft.com/excel/stop-automatically-changing-numbers-to-dates)). Since 2023, Excel has had options under File > Options > Data > Automatic Data Conversion to stop removing leading zeros, cutting numbers to 15 digits, reading numbers around an "E" as scientific notation and turning some letter-and-number strings into dates ([Microsoft 365 Insider blog](https://techcommunity.microsoft.com/blog/microsoft365insiderblog/control-data-conversions-in-excel-for-windows-and-mac/4215336)). This bites hardest when you open a CSV, because every value goes through those rules. Segments that are only a part number or a version are the usual victims.

**Encoding.** Microsoft says a UTF-8 CSV opens normally only if it was saved with a byte order mark; without one, you import it through Data > Get Data > From Text/CSV ([Microsoft](https://support.microsoft.com/en-us/office/opening-csv-utf-8-files-correctly-in-excel-8a935af5-3416-4edd-ba7e-3dfd2bc4a032)). Otherwise "traducción" turns into mojibake, which is a lovely word for a sad thing. The CSVs TraduQ writes carry that mark, and an .xlsx doesn't have the problem at all.

**Size limits.** A cell holds at most 32,767 characters and a worksheet 1,048,576 rows ([Microsoft](https://support.microsoft.com/en-us/office/excel-specifications-and-limits-1672b34d-7043-467e-8e27-269d656771c3)). Segments that long are rare, but memories built from badly segmented files have them. If a memory has more units than rows, use CSV or split it first with [TMX split](https://traduq.com/tools/tmx-split).

**The separator.** Excel in Spanish and most of Europe expects semicolons in CSV, not commas. That's why the converter lets you choose. Another reason I prefer .xlsx.

## Other ways to do it

**memoQ.** memoQ exports memories as TMX (its docs describe [the server export](https://docs-memoq-com.azurewebsites.net/current/en/Workspace/export-translation-memories-on.html), one file per memory), and it can import CSV or tab-separated files into a translation memory, mapping each column to the source, the target or fields like Client, Project, Domain or Author ([memoQ docs](https://docs-memoq-com.azurewebsites.net/current/en/Workspace/translation-memory-csv-import.html)). So the way back can keep some metadata, if you kept those columns in the sheet.

**Trados Studio.** Studio exports a memory from the Translation Memories view with Home > Export, optionally filtered, and the RWS docs say ".tmx" is the standard format for exported TMs ([RWS](https://docs.rws.com/en-US/trados-studio-2024-1145319/exporting-tm-data-361163)). For the spreadsheet step you need a converter.

**Maxprograms TMXEditor.** A desktop editor for Windows, macOS and Linux. Its product page lists conversion from Excel and CSV or tab-delimited files to TMX, and export to Excel and tab-delimited files ([Maxprograms](https://maxprograms.com/products/tmxeditor.html)). Its tab-delimited export uses Excel's defaults, UTF-16LE with tabs and no text delimiters ([manual](https://maxprograms.com/manuals/tmxeditor/en/export_delimited.html)). The installer needs a subscription after a 7-day trial; the source code is open.

**Olifant.** The Okapi project's TMX editor can import tab-delimited tables, expects them in UTF-8 and lets you assign a language to each column ([Olifant help](https://okapi.sourceforge.net/Release/Olifant/Help/ui_tableimport.htm)). The Okapi wiki says the Java version is under development and points Windows users to the older .NET version ([Okapi wiki](https://okapiframework.org/wiki/index.php/Olifant)).

## When not to use a spreadsheet

If you only need to fix a handful of units, skip Excel. The [TMX editor](https://traduq.com/tools/tmx-editor) opens the memory itself, shows tags as numbered chips you can't break by accident, and saves a TMX with the original tag XML. To remove duplicates, empty units or conflicting translations across the whole memory, the [TM cleaner](https://traduq.com/tools/tm-cleaner) does it while keeping the header, properties and dates byte for byte. I wrote up both in [how to clean a TMX translation memory](https://traduq.com/blog/clean-tmx-translation-memory). And if you're putting several memories together, see [how to merge TMX files](https://traduq.com/blog/merge-tmx-files) first.

The spreadsheet wins when a human who doesn't use CAT tools has to read thousands of units, or when you want Excel's filters and sorting on the text.

## Common problems

**The Spanish column is full of strange characters.** The file was read in the wrong encoding. Use the .xlsx, or open the CSV through Data > Get Data instead of double-clicking it.

**Tags came back as text.** Someone edited the tag code in the cell. Find the unit, copy the tags from the source cell or the original TMX, and convert again.

**One segment is split over two rows.** Someone pasted text with line breaks onto the sheet. Put the text back into one cell, with Alt+Enter where the break should be.

**"Add a language code".** A column has no language in its header. Type the code before converting, or fix the header in the sheet.

**My CAT tool shows the units as new.** That's expected. The round-tripped TMX has new units with no dates or users. Import it into the existing memory with the settings you want, and keep the original.

## Frequently asked questions

### How do I open a TMX file in Excel?

Convert it first. Drop the TMX in the [TMX to Excel converter](https://traduq.com/tools/tmx-converter), choose Excel (.xlsx) and download. Each language becomes a column with its code in the header row.

### Can I convert an Excel file into a TMX?

Yes. Put each language in its own column with a code such as en-US or es-MX in the first row, drop the .xlsx in the converter and choose TMX.

### Are inline tags kept?

Yes, if you choose "Keep" and nobody edits the tag code. They appear in the cells as XML and become real tags again in the TMX.

### Does it work with a multilingual TMX?

Yes. Every language becomes a column, and you can convert some or all of them back into one TMX.

### Are properties, notes and dates kept?

No. A spreadsheet has no place for them, so the TMX you get back has text, tags and unit IDs only. Keep the original memory for the metadata.

### Is my memory uploaded?

No. The conversion runs in your browser and the file never reaches TraduQ's server, which matters for memories under an NDA.

I'm a freelance EN↔ES translator, and I built TraduQ as a marketplace where clients find translators. Its file tools are free, need no sign-up and run in your browser. For what TMX, XLIFF and TBX each hold, see [XLIFF, TMX and TBX explained](https://traduq.com/blog/xliff-tmx-tbx-formats-explained).

## Sources

Pages checked on 17 November 2026:

- [GALA: TMX 1.4b Specification](https://www.gala-global.org/tmx-14b): structural and inline elements
- [Microsoft: Excel specifications and limits](https://support.microsoft.com/en-us/office/excel-specifications-and-limits-1672b34d-7043-467e-8e27-269d656771c3): 32,767 characters per cell, 1,048,576 rows
- [Microsoft: Stop automatically changing numbers to dates](https://support.microsoft.com/excel/stop-automatically-changing-numbers-to-dates)
- [Microsoft 365 Insider: Control data conversions in Excel for Windows and Mac](https://techcommunity.microsoft.com/blog/microsoft365insiderblog/control-data-conversions-in-excel-for-windows-and-mac/4215336)
- [Microsoft: Opening CSV UTF-8 files correctly in Excel](https://support.microsoft.com/en-us/office/opening-csv-utf-8-files-correctly-in-excel-8a935af5-3416-4edd-ba7e-3dfd2bc4a032)
- [Microsoft: Start a new line of text inside a cell in Excel](https://support.microsoft.com/en-us/office/start-a-new-line-of-text-inside-a-cell-in-excel-33e41eab-8b5e-4193-93d6-9a06ecf812b3)
- [memoQ: Export translation memories on server](https://docs-memoq-com.azurewebsites.net/current/en/Workspace/export-translation-memories-on.html)
- [memoQ: Translation memory CSV import settings](https://docs-memoq-com.azurewebsites.net/current/en/Workspace/translation-memory-csv-import.html)
- [RWS: Exporting TM data (Trados Studio 2024)](https://docs.rws.com/en-US/trados-studio-2024-1145319/exporting-tm-data-361163)
- [Maxprograms: TMXEditor](https://maxprograms.com/products/tmxeditor.html) and [Export as TAB delimited file](https://maxprograms.com/manuals/tmxeditor/en/export_delimited.html)
- [Okapi: Olifant table import options](https://okapi.sourceforge.net/Release/Olifant/Help/ui_tableimport.htm) and [Olifant on the Okapi wiki](https://okapiframework.org/wiki/index.php/Olifant)

## About the author

[Andrés Lacruz](https://traduq.com/translators/andres-lacruz-6ae1gbuo) — EN↔ES translator and creator of TraduQ

## Related posts

- [How to Clean a TMX Translation Memory (Free Tool)](https://traduq.com/blog/clean-tmx-translation-memory) — Remove duplicates, empty and broken units, extra spaces and conflicting translations from a TMX file, step by step, and review every change first.
- [How to Open an SDLTM File Without Trados (Free)](https://traduq.com/blog/open-sdltm-without-trados) — Open a Trados Studio .sdltm translation memory without a licence and convert it to TMX, Excel or CSV in your browser. Step by step, plus other ways.
- [XLIFF, TMX, and TBX: Translation File Formats Explained](https://traduq.com/blog/xliff-tmx-tbx-formats-explained) — What is inside an XLIFF, SDLXLIFF, MQXLIFF, TMX or TBX file, which tool made it, and how to open, edit or convert it for free without Trados or memoQ.

## Tools for this

- [TMX and XLIFF converter](https://traduq.com/tools/tmx-converter) — TMX, XLIFF, SDLXLIFF and PO to Excel or CSV, and back.
- [TMX editor](https://traduq.com/tools/tmx-editor) — Edit target segments with memoQ-style tags, delete units and save a TMX.
- [TM cleaner](https://traduq.com/tools/tm-cleaner) — Find and remove duplicates, empty and broken segments in a TMX.

## Glossary

- [Fuzzy match](https://traduq.com/glossary/fuzzy-match)
- [SDLXLIFF](https://traduq.com/glossary/sdlxliff)
- [TMX](https://traduq.com/glossary/tmx)
- [Translation memory](https://traduq.com/glossary/translation-memory)
- [XLIFF](https://traduq.com/glossary/xliff)
