---
title: "CSV contains invalid characters"
description: "Fix the invalid-character error when you import a CSV. Clean the header row, save the file as UTF-8, and upload it again."
sidebar:
  label: "CSV invalid characters"
---

If import shows an invalid-character error, the header row of your CSV contains broken characters. Save the file as UTF-8 with a clean header row and upload it again.

The error appears after you click `Save`, in the `Processing failed` dialog:

> The CSV file contains invalid characters. Please save it as UTF-8 and try again.

<img src="/_assets/importing-data/troubleshoot-csv-encoding--processing-failed.png" alt="Processing failed dialog over the Import data page, with the invalid-characters message and a Try again button" width="1600" height="960" />

## Most likely cause: the header row contains replacement characters

MangoMap checks the header row (the first line) for the replacement character `�`. This character is left behind when an earlier tool could not read the text and saved the file anyway. MangoMap cannot get the original characters back.

**How to verify.**

Open the file in a text editor and look at the first line. If you see `�` in a column name, you have found the cause.

**Fix.**

Retype the damaged column names in the first line, then save the file as UTF-8 and upload it again. If the same character is in the data rows, go back to the original source of the data and export it again as UTF-8.

## Related problem: the file is in another encoding

MangoMap reads UTF-8 files. If a file is not valid UTF-8, MangoMap reads it as Windows-1252 (a common Excel default on Windows). This works for Windows-1252 files, but other encodings give a different symptom, not the invalid-character error:

| Encoding | What you see |
| --- | --- |
| UTF-16 (`UTF-16 LE`, `UTF-16 BE`, Excel `Unicode Text`) | Processing fails with `Your CSV is missing a longitude column`, although the column is in the file. |
| ISO-8859-2, Shift-JIS, and other non-Windows-1252 encodings | The import finishes, but accented or non-Latin text shows as the wrong characters. |
| UTF-8 with a few stray bytes near the top, from a copy-paste or a database export | The import finishes, but accented text shows as the wrong characters (for example, `é` for `é`). |

**How to verify.**

Open the file in a text editor that shows the encoding (VS Code shows the encoding in the bottom-right). If it shows an encoding other than `UTF-8` or `Windows-1252`, you have found the cause.

**Fix.**

Save the file again as UTF-8:

- **Excel (Windows or macOS):** `File → Save As → CSV UTF-8 (Comma delimited) (*.csv)`. Don't use plain `CSV` — that uses the system encoding, not UTF-8.
- **Numbers (macOS):** `File → Export To → CSV → Advanced Options → Text Encoding: Unicode (UTF-8)`.
- **LibreOffice Calc:** `File → Save As`, check `Edit filter settings`, then choose character set `Unicode (UTF-8)`.
- **Google Sheets:** `File → Download → Comma-separated values (.csv)` — Google Sheets always exports UTF-8.
- **Command line:** run `iconv`. Replace `UTF-16` with the encoding your text editor shows.

  ```console
  $ iconv -f UTF-16 -t UTF-8 input.csv > output.csv
  ```

Upload the new file. If you already imported the file with the wrong characters, delete that dataset and import the new file.

## If none of the above

Open the file in any UTF-8-aware editor (VS Code, Sublime Text), copy all the contents, paste into a new file, and save as `data-fixed.csv` with UTF-8 encoding. This creates a clean UTF-8 file.

## Related

- [How to import a CSV](/docs/importing-data/import-csv)
- [CSV missing longitude or latitude column](/docs/importing-data/troubleshoot-csv-missing-coordinates)