I'm writing a filename I/O procedure in x86-16 assembly language. It takes eight characters (I don't need to support long filenames) from the keyboard and prints them to an on-screen text input field.

At the moment I'm allowing numbers, upper/lower-case letters, underscores, and hyphens.

I'd like to allow all legal symbols, but I can't find an official list of banned characters. Common sense tells me that slashes are illegal, but if I had to guess, I would say that the plus character is legal. (edit: It's not!)

I'm already ignoring the period character since my code automatically handles appending the period and file extension.

  • 17
    You might also find Retrocomputing useful. – Bob Sep 28 '18 at 7:27
  • Try to create a folder in Windows and put a '?' in the name. A tooltip tells you which characters are forbidden. This gives you a start :)... – Mixxiphoid Sep 28 '18 at 9:49
  • @Mixxiphoid that won't work because the set of allowed characters in Windows are much larger. For example +,;[], space and a-z are allowed in Windows but not DOS. Explorer gives me the error "A file name can't contain any of the following characters \ / : * ? " < > | which is just a subset of the banned characters in DOS – phuclv Sep 28 '18 at 14:44
  • 1
    @phuclv that is why I said 'This gives you a start' and also why this is a comment and not an answer. – Mixxiphoid Sep 28 '18 at 14:46
  • Why all MS-DOS symbols? Why not also consider other older OS rules? – jpmc26 Sep 30 '18 at 0:45

A concise summary can be found on Wikipedia:

Legal characters for DOS filenames include the following:

  • Upper case letters AZ
  • Numbers 09
  • Space (though trailing spaces in either the base name or the extension are considered to be padding and not a part of the filename, also filenames with spaces in them must be enclosed in quotes to be used on a DOS command line, and if the DOS command is built programmatically, the filename must be enclosed in quadruple quotes when viewed as a variable within the program building the DOS command.)
  • ! # $ % & ' ( ) - @ ^ _ ` { } ~
  • Values 128–255 (though if NLS services are active in DOS, some characters interpreted as lowercase are invalid and unavailable)

This excludes the following ASCII characters:

  • " * + , / : ; < = > ? \ [ ] | [9]
  • Windows/MS-DOS has no shell escape character
  • . (U+002E . full stop) within name and extension fields, except in . and .. entries (see below)
  • Lower case letters az (stored as A–Z on FAT12/FAT16)
  • Control characters 0–31
  • Value 127 (DEL)[dubious – discuss]


And here's what MS-DOS 6 user guide officially said

Naming Files and Directories

Every file and directory, except for the root directory on each drive, must have a name. The following list summarizes the rules for naming files and directories. File and directory names:

  • Can be up to eight characters long. In addition, you can include an extension up to three characters long.
  • Are not case-sensitive. It does not matter whether you use uppercase or lowercase letters when you type them.
  • Can contain only the letters A through Z, the numbers 0 through 9, and the following special characters: underscore (_), caret (^), dollar sign ($), tilde (~), exclamation point (!), number sign (#), percent sign (%), ampersand (&), hyphen (-), braces ({}), at sign (@), single quotation mark (`), apostrophe ('), and parentheses (). No other special characters are acceptable.
  • Cannot contain spaces, commas, backslashes, or periods (except the period that separates the name from the extension).
  • Cannot be identical to the name of another file or subdirectory in the same directory.

This is from PC-DOS 7:

The name you assign to a file must meet the following criteria:

  • It can contain no more than eight characters.
  • It can consist of the letters A through Z, the numbers 0 through 9, and the following special characters:

    _ underscore            ^  caret
    $ dollar sign           ~  tilde
    ! exclamation point     #  number sign
    % percent sign          &  ampersand
    - hyphen                {} braces
    @ at sign               `  single quote
    ' apostrophe            () parentheses

Note: No other special characters are acceptable.

  • The name cannot contain spaces, commas, backslashes, or periods (except the period that separates the name from the extension).
  • The name cannot be one of the following reserved file names: CLOCK$, CON, AUX, COM1, COM2, COM3, COM4, LPT1, LPT2, LPT3, LPT4, NUL, and PRN.
  • It cannot be the same name as another file within the directory.

User's Guide - PC DOS 7

The first byte of a name must not be 0x20 (space). Short names or extensions are padded with spaces. Special ASCII characters 0x22 ("), 0x2a (*), 0x2b (+), 0x2c (,), 0x2e (.), 0x2f (/), 0x3a (:), 0x3b (;), 0x3c (<), 0x3d (=), 0x3e (>), 0x3f (?), 0x5b ([), 0x5c (\), 0x5d (]), 0x7c (|) are not allowed.

The FAT filesystem

If you're also interested in MS-DOS 5.0 then here it is.

  • 11
    It might be worth noting that even though they only contain valid characters the special filenames CON, PRN, AUX, NUL, COM1, COM2, COM3, COM4, COM5, COM6, COM7, COM8, COM9, LPT1, LPT2, LPT3, LPT4, LPT5, LPT6, LPT7, LPT8, and LPT9 are also not permitted (see here) – Bill Tür Sep 28 '18 at 10:42
  • 3
    @ThomasSchremser "Do not use", "Avoid" and "Not recommended" is not the same as "not permitted". – RobIII Sep 28 '18 at 14:23
  • 1
    @RobIII Yes but they linked to "Windows > Desktop" documentation, not to "MS-DOS" documentation. The wiki for DOS says, "There are reserved device names in DOS that cannot be used as filenames regardless of extension as they are occupied by built-in character devices". In other words, not permitted in DOS and some versions of Windows, and not recommended in other Windows versions. – Quantic Sep 28 '18 at 18:31
  • It's interesting that the ` is referred to as a single quote. I've always heard it called a backtick, and the ' (what they call (not incorrectly) an apostrophe) as a single quote. – ale10ander Sep 28 '18 at 18:41
  • 2
    @ale10ander yeah that surprised me to. I've always hated that many people use it for the apostrophe (like in I`m) or the begin part of the quote. For example GNU documentations always write `like this' which is very ugly and less readable to me – phuclv Sep 28 '18 at 19:06

Strictly speaking, as an MS/PC/DR-DOS applications programmer you are supposed to ask the operating system for this information. INT 0x21 with AX=0x6505 returns a pointer to the so-called FCHAR NLS table for your country and code page. This table lists a range of characters and a further set of characters that terminate filenames.

In theory it varies by country and code page. But the fact that it was not formally carried over into the OS/2 Control Program API and the fact that FreeDOS has 1 table across all codepages and countries show that it is largely invariant in practice.

Further reading


I found this in a manual for MS-DOS 3.3. I'm running 6.22, but it probably still applies. I was wrong about '+' being allowed.

Enter image description here

  • 2
    A manual from back-in-the-day is more reliable than Wikipedia – Stewart Sep 28 '18 at 12:53
  • @Stewart what's important are the quotes on Wikipedia, not Wikipedia itself. If in doubt just check the footnotes and references in the article \@Mylifeisabug I've just added the MS-DOS 6 manual – phuclv Sep 28 '18 at 15:52

If you just want to validate the filename, you may want to use INT 21H/AH=60H (TRUENAME - CANONICALIZE FILENAME OR PATH) after ensuring that the passed filename doesn't have a colon or backslash (those may be treated as drive letters and directories): the function takes your proposed filename and tries to canonicalize it by uppercasing the letters and checking for invalid characters (it also adds a drive letter/server name and path.)

In pseudocode:

If !(filename contains {"/", "\", ".", ":"})
    Canonicalize filename (INT 21H/AH=60H)
    If !(CF is set) filename is valid
Filename is not valid

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.