📄 CSV Fundamentals · 7 min read

CSV vs TXT: What's the Difference?

Both are plain text, and the bytes are often identical. The difference is the promise the extension makes: .csv means records of comma-separated fields, registered as text/csvin RFC 4180, while .txt means only "this is text" with no claim about fields at all. That promise changes behaviour. Microsoft's own documentation says a .txt file starts the Import Text Wizard, and a .csv file opens straight into a workbook with the default data format applied to every column.

So the useful question is not which format is "better". It is which one keeps the receiving program from guessing wrong, and that depends on whether your file is actually a table.

What each extension promises

RFC 4180 gives CSV a real grammar. Section 2 defines the file as file = [header CRLF] record *(CRLF record) [CRLF] with record = field *(COMMA field): optional header line, then records, fields separated by the comma character, records ending in CRLF. It also allows an optional header parameter in the media type registration whose values are present or absent, so a CSV can even declare whether its first line is column names.

A .txt file declares none of that. It is plain text: the bytes are characters, line breaks are line breaks, and whether a comma or a tab means "next column" is entirely the reader's problem. This is not a weakness — it is the point of a text file. But it explains why everything that has to parse data treats the two extensions differently.

What Excel actually does with each one

Microsoft's support article on importing and exporting text files is explicit about the split, and worth quoting because it settles most arguments:

"If the file is a text file (.txt), Excel starts the Import Text Wizard... If the file is a .csv file, Excel automatically opens the text file and displays the data in a new workbook."

The same article adds the part people rarely read: "When Excel opens a .csv file, it uses the current default data format settings to interpret how to import each column of data." No prompt, no preview, no choice of encoding. Whatever your regional settings think a date looks like is what you get.

The article then gives the workaround that tells you how mechanical all of this is: "To force Excel to run the Import Text Wizard, you can change the file name extension from .csv to .txt before you open it." The contents did not change. The route did.

CSV vs TXT side by side

What we measured

We wrote one 40-byte file — two records plus CRLF line endings — three times, changing only the extension, then asked Python's mimetypes module what each one is. On 2026-10-31, Python 3.13.5:

Three different types, one identical file. That is the whole mechanism of this article in a table: the extension is a label, and downstream software reads the label before it reads the bytes.

Two follow-up observations from the same session. First, a library that parses content rather than extensions does not care: pandas.read_csv() returned the same three-column frame from all three files, because it reads bytes and assumes a comma. Second, the label wins over the truth: we took a tab-separated file and named it tabs-named.csv. pandas, still assuming a comma, returned one column named id followed by a tab character and the word name. Passing sep="\t" returned two columns. Nothing was corrupt; the file simply was not the thing its name said it was.

When TXT is the better choice

A .txt is the right container when there is no table. If the file is a log with one event per line, a list of emails, a SQL dump you intend to pipe into something else, or notes that happen to contain commas, calling it .csv invites every importer on the machine to invent columns that do not exist.

It is also the honest choice for a file whose delimiter you want a human to choose. That is why the wizard exists. Tab-delimited text with a .txt extension says "this is text, work out the structure yourself" — which is exactly the behaviour you want when you are about to check every column type by hand.

When renaming is the wrong fix

Renaming fixes routing, not content, and it has three traps.

Hidden extensions. Windows hides extensions by default, so a file displayed as data.csv may really be data.csv.txt. Turn on View > Show > File name extensions before you rename anything, or you will end up with two extensions and a file that still will not open where you expect. Why your CSV behaves like a text file covers that failure in detail.

The default format problem. A .csv that opens straight into Excel gets each column interpreted with your regional defaults. Leading zeros disappear, long numbers turn into scientific notation, and day-first dates may swap. None of this happens in the file; it happens on the way in. If any of those columns matter, rename to .txt first and set the types in the wizard, or import through Data > From Text/CSV.

Delimiter mismatch. If the fields are separated by semicolons or tabs, renaming to .csv produces a file that claims commas and contains none. Fix the delimiter itself: the delimiter converter rewrites the separators with proper quoting, which is safer than find-and-replace on any file where values contain commas. The delimiter options guide explains which separator a given export tends to use.

Converting between the two

Going from a real CSV to a tab-delimited .txt is a delimiter change, not a format change: same rows, same order, different separator. The converter does that in the browser and quotes fields that need it. Going the other way — a delimited text file into CSV — is the same operation with the separator set to tab. Converting text to CSV walks through it, and the RFC 4180 walkthrough explains what a valid CSV must contain.

One habit that prevents most of these problems: decide the extension from the content before the file leaves your machine. If a system exports tab-separated data as export.csv, convert it once, at the source of the confusion, instead of explaining the mismatch to every person who opens it later.

Frequently Asked Questions

Sources

• RFC 4180, Common Format and MIME Type for CSV Files, section 2 (ABNF grammar: optional header, records separated by CRLF, fields separated by COMMA) and section 3 (registration of text/csv under the text media type; optional charset and headerparameters with values present or absent; "Common usage of CSV is US-ASCII"). Retrieved 2026-10-31.
• Microsoft Support, Import or export text (.txt or .csv) files: delimited text files (.txt) typically use tab (ASCII 009) and CSV files use the comma; a .txt opened in Excel starts the Import Text Wizard, a .csv is opened automatically into a new workbook using the current default data format settings; to force the wizard on a CSV, change the extension to .txt; import and export are supported up to 1,048,576 rows and 16,384 columns. Retrieved 2026-10-31.
• Local measurement, 2026-10-31, Python 3.13.5: one 40-byte file written as data.csv, data.txt and data.tsv; mimetypes.guess_type returned text/csv, text/plain and text/tab-separated-values; pandas 2.3.3 read all three as shape (2, 3) with columns id, name, city; a tab-separated file named tabs-named.csv read as shape (1, 1) with the single column id\tname, and as shape (1, 2) with sep="\t".