ASCII Table

What ASCII is

ASCII is a table of 128 codes, from 0 to 127, using 7 bits per character. It started in the 1960s as an American standard (ANSI X3.4) and is described in RFC 20. It holds the unaccented letters, the digits, punctuation and 33 control characters that never show up on screen.

Anything above 127 is no longer ASCII, even when someone calls it "extended ASCII".

Control characters you still meet every day

Most of the 33 control characters fell out of use, but a few turn up every week. NUL (0) ends a string in C and shows up in binary files. TAB (9) is the tab. LF (10) and CR (13) make the line break: Linux and macOS use LF alone, Windows uses the pair CR and LF (CRLF). ESC (27) starts the terminal color sequences, and DEL (127) is the only control outside the 0 to 31 range.

The line break difference causes real bugs, like a shell script that will not run because of a CR at the end of the line. Git can convert on commit and checkout, depending on core.autocrlf: true on Windows and input on Mac and Linux usually fixes it. In CSV, a line break inside a field is only read correctly when the field is quoted.

git config --global core.autocrlf input

There is no single "extended ASCII"

When people say "extended ASCII" they almost always mean one of three tables: ISO-8859-1, Windows-1252 or CP850. They match from 0 to 127 and differ from there up. And the browser does not even treat "latin1" as ISO-8859-1: under the WHATWG Encoding Standard, "latin1", "iso-8859-1" and even "ascii" are read as Windows-1252.

ISO-8859-1 (from 1987) uses 128 to 159 for C1 controls, which nobody types. Windows-1252 puts useful symbols in those 32 codes, like €, “ ”, … and •, and leaves only five empty. CP850 is something else: it is the code page of the Windows console and of Alt codes, and there "ç" is 135, not 231.

When you need one of these tables, say its name. "ASCII" only goes up to 127.

ASCII, Unicode and UTF-8

Unicode is the list of all characters, and UTF-8 is the most common way to store them as bytes. From 0 to 127, UTF-8 uses one byte with the same value as ASCII, so a file with only ASCII is identical in both. Above that, each character takes 2 to 4 bytes: an accented letter takes 2, a symbol like € takes 3 and an emoji takes 4.

That matters when the limit is in bytes, like a database column or a header. The Character and Word Counter shows both numbers side by side. In JavaScript, note that length counts code units, not characters or bytes:

"ç".length                             // 1
new TextEncoder().encode("ç").length   // 2
"😀".length                            // 2
new TextEncoder().encode("😀").length  // 4

Why "ç" turns into "ç"

"ç" in UTF-8 takes two bytes, C3 and A7. If a program reads those bytes as Windows-1252, each byte becomes its own character: C3 is "Ã" and A7 is "§". That is where "ç" comes from. Almost always one end is set up wrong: the database in latin1, the CSV saved without a BOM, the Content-Type header without a charset, or the file saved as "ANSI".

The usual suspects:

• A MySQL table in latin1 instead of utf8mb4. Data goes in right and comes out wrong, or the other way around.
• A UTF-8 CSV opened in Excel without a BOM. Excel assumes the Windows code page and shows mojibake.
• An HTTP response without charset in the Content-Type header. The browser guesses the encoding.
• A file saved as "ANSI", which on Western European and Brazilian Windows is Windows-1252.
• btoa only takes characters up to 255. btoa("€") throws InvalidCharacterError, and btoa("ç") works but encodes the byte E7 (latin1), not C3 A7 (UTF-8). For Base64 of UTF-8, convert to bytes first. The Base64 Decoder does that.

btoa("€");                                              // InvalidCharacterError
btoa("ç");                                              // "5w==" (byte E7)
btoa(String.fromCharCode(...new TextEncoder().encode("ç"))); // "w6c=" (bytes C3 A7)

ALTER TABLE clientes CONVERT TO CHARACTER SET utf8mb4;

Hashes have the same problem: the SHA-256 of "ç" is not the same in latin1 (byte E7) and in UTF-8 (bytes C3 A7). If a hash does not match the other system's, check the text encoding in the Hash Generator.

ASCII in code

The table explains tricks that show up in all kinds of code. Uppercase and lowercase letters differ by 32 (a single bit), and the digits "0" to "9" sit at 48 to 57, so c - '0' in C or Java turns the character into a number. In regex, [ -~] matches printable characters, from space (32) to tilde (126).

"a".charCodeAt(0) - "A".charCodeAt(0); // 32
"7".charCodeAt(0) - "0".charCodeAt(0); // 7
/^[ -~]*$/.test("Olá");                // false
"B".charCodeAt(0) - 48;                // 18

The alphanumeric CNPJ uses the same idea. To compute the check digit, each character is worth its ASCII code minus 48: "0" to "9" still count 0 to 9, "A" is 17, "B" is 18 and so on up to "Z", which is 42. To see it working, use the CNPJ Generator or check a number in the CNPJ Validator.

HTML entities and escapes

In HTML you only need an entity for characters the parser reads as markup: &, < and >, plus " inside attributes.   is the non-breaking space, code 160, and it is not the same as a regular space (32).

With the page in UTF-8, accents and symbols can go straight into the HTML: you do not need ç or ç. In JavaScript and JSON, the \u00e7 escape helps when the file is not UTF-8, and \n, \t and \ are the short forms. In a URL, "ç" becomes %C3%A7, which are its UTF-8 bytes.

Alt codes on Windows

Alt codes are not ASCII. Alt+NNN, without a zero, uses CP850 on Windows in Portuguese, Spanish, French and German: Alt+135 types "ç". Alt+0NNN, with a leading zero, uses Windows-1252: Alt+0231 types the same "ç". The number changes because the table changes.

Always type on the numeric keypad, with Num Lock on. On a laptop without a keypad, the easiest way is to copy the character from the table above.

Sources

Checked on November 3, 2026:

• RFC 20, ASCII format for Network Interchange
• WHATWG Encoding Standard
• CP1252 to Unicode mapping (Microsoft, on unicode.org)
• CP850 to Unicode mapping (Microsoft, on unicode.org)
• Microsoft: insert ASCII or Unicode Latin-based symbols and characters