ASCII Table

ASCII table — all 128 codes in decimal, hex, octal and binary

The complete ASCII table: 33 control codes named and explained, 95 printable characters, and every code shown four ways at once. Type a character to get its codes, or type a number in any base to get the character back. It also settles the question that brings most people here — there is no such thing as “extended ASCII”, and the table further down shows three encodings disagreeing about the same byte. Nothing is uploaded or saved.

Nothing is uploaded or saved
Works offline

Look up a character or a code

One box, both directions. Paste a character, or type a number as 65, 0x41, \x41, U+0041, 0o101 or 0b1000001.

Matched by the character “A” — LATIN CAPITAL LETTER A

Decimal

Hex

0x41

Octal

0101

Binary

01000001

Character

JS escape

HTML

CSS / URL

\41 · %41

Keyboard

—

Hex 41. The capitals occupy 65–90 with no gaps, and the lowercase run starts exactly 32 higher — so changing case is a single bit, which is why old code did it with an OR of 0x20.

The full table — 128 codes

128 of 128 shown

Dec	Hex	Oct	Binary	Char	Name	Description
0	00	000	00000000	␀	NUL · Null	[Ctrl+@ \0]All bits zero. On paper tape it punched no holes at all, which made it usable as filler and let it be over-punched into any other code later. Today it ends a string in C, and it is the byte that most often breaks a filename or a database field.still used
1	01	001	00000001	␁	SOH · Start of Heading	[Ctrl+A]Marked the start of a message’s addressing header, before the text itself began.
2	02	002	00000010	␂	STX · Start of Text	[Ctrl+B]Ended the heading and began the message body.
3	03	003	00000011	␃	ETX · End of Text	[Ctrl+C]Closed the message body. It survives as Ctrl+C, which Unix terminals repurposed into the interrupt key.
4	04	004	00000100	␄	EOT · End of Transmission	[Ctrl+D]Told the receiver the sender had finished and the line was free. It survives as Ctrl+D, which signals end of input on Unix.
5	05	005	00000101	␅	ENQ · Enquiry	[Ctrl+E]Asked the far end “who are you?” and expected an identification in reply.
6	06	006	00000110	␆	ACK · Acknowledge	[Ctrl+F]Positive reply: the last block arrived intact, send the next one.
7	07	007	00000111	␇	BEL · Bell	[Ctrl+G \a]Rang a physical bell on the teleprinter to get a human’s attention. Terminals still beep on it, and \a is still the escape for it in C.still used
8	08	010	00001000	␈	BS · Backspace	[Ctrl+H \b]Moved the carriage back one position so the next character struck over the last. That is how you produced a bold or an accented letter on paper.still used
9	09	011	00001001	␉	HT · Horizontal Tab	[Ctrl+I \t]Advanced to the next tab stop. Still everywhere: \t in every language, and the column separator in TSV files.still used
10	0A	012	00001010	␊	LF · Line Feed	[Ctrl+J \n]Advanced the paper by one line without moving the carriage sideways. Unix, macOS and effectively every file format designed since use it alone as the newline.still used
11	0B	013	00001011	␋	VT · Vertical Tab	[Ctrl+K \v]Advanced to the next vertical tab stop, for pre-printed forms. Almost nothing reaches for it now.
12	0C	014	00001100	␌	FF · Form Feed	[Ctrl+L \f]Ejected the page and started a new one. Printers still honour it, and Ctrl+L clears a terminal screen for exactly that reason.still used
13	0D	015	00001101	␍	CR · Carriage Return	[Ctrl+M \r]Returned the print head to column 1 without advancing the paper. Windows still writes CR LF as a pair, and HTTP, SMTP and the other line-based internet protocols require the pair.still used
14	0E	016	00001110	␎	SO · Shift Out	[Ctrl+N]Switched the device to an alternate character set, which stayed in effect until SI switched it back.
15	0F	017	00001111	␏	SI · Shift In	[Ctrl+O]Switched back to the standard character set.
16	10	020	00010000	␐	DLE · Data Link Escape	[Ctrl+P]Made the character following it count as a control code again, inside a stream that was otherwise being carried as raw data.
17	11	021	00010001	␑	DC1 · Device Control 1	[Ctrl+Q]Left to the device by the standard, but by near-universal convention it became XON: resume sending. Ctrl+Q is what unfreezes a terminal.still used
18	12	022	00010010	␒	DC2 · Device Control 2	[Ctrl+R]Device-specific with no agreed meaning — terminals and printers each assigned it differently.
19	13	023	00010011	␓	DC3 · Device Control 3	[Ctrl+S]By the same convention, XOFF: stop sending. Ctrl+S is why a terminal sometimes appears to freeze for no reason at all.still used
20	14	024	00010100	␔	DC4 · Device Control 4	[Ctrl+T]Device-specific, often used as a general stop, but with no single agreed meaning across devices.
21	15	025	00010101	␕	NAK · Negative Acknowledge	[Ctrl+U]The negative counterpart of ACK: the last block was damaged, send it again.
22	16	026	00010110	␖	SYN · Synchronous Idle	[Ctrl+V]Sent to fill a synchronous line that had no data on it, so the two ends stayed in step.
23	17	027	00010111	␗	ETB · End of Transmission Block	[Ctrl+W]Ended one block of a message that had been split for transmission, where ETX ended the whole message.
24	18	030	00011000	␘	CAN · Cancel	[Ctrl+X]Told the receiver that the data preceding it was in error and should be discarded.
25	19	031	00011001	␙	EM · End of Medium	[Ctrl+Y]Marked the physical end of the tape, card or roll of paper — which is not the same thing as the end of the data.
26	1A	032	00011010	␚	SUB · Substitute	[Ctrl+Z]Stood in for a character that arrived damaged. CP/M and then DOS used Ctrl+Z to mark the end of a text file, which is why it still ends console input on Windows.still used
27	1B	033	00011011	␛	ESC · Escape	[Ctrl+[]Introduced a sequence meaning something other than itself. Every ANSI terminal sequence starts with this byte, so the colour in your terminal begins at code 27. Shells and GNU C accept \e for it, but that is an extension — standard C and JavaScript have no shorthand, so you write \x1b or \u001b.still used
28	1C	034	00011100	␜	FS · File Separator	[Ctrl+\]The coarsest of four nested separators — file, group, record, unit — meant for structuring data without inventing a delimiter that might appear in the data itself.
29	1D	035	00011101	␝	GS · Group Separator	[Ctrl+]]One level in from FS. Still genuinely used: GS1 barcodes separate their application-identifier fields with it.still used
30	1E	036	00011110	␞	RS · Record Separator	[Ctrl+^]One level in from GS. JSON Text Sequences (RFC 7464) put it before each record, which is how a stream of JSON documents stays unambiguous.still used
31	1F	037	00011111	␟	US · Unit Separator	[Ctrl+_]The finest of the four, sitting between individual fields of one record.
32	20	040	00100000	␠	SPACE	A real character with a code, not the absence of one. It becomes %20 in a URL path and, in form-encoded query strings only, a “+”.
33	21	041	00100001	!	EXCLAMATION MARK
34	22	042	00100010	"	QUOTATION MARK	["]The straight typewriter quote. The curly “ ” pair your word processor substitutes is not ASCII and is a frequent cause of code that will not parse.
35	23	043	00100011	#	NUMBER SIGN	Hash, pound or octothorpe depending on who you ask. It starts the fragment in a URL, and everything after it is never sent to the server.
36	24	044	00100100	$	DOLLAR SIGN
37	25	045	00100101	%	PERCENT SIGN	Begins a percent-escape in a URL, so a literal one has to be written %25.
38	26	046	00100110	&	AMPERSAND	[&]
39	27	047	00100111	'	APOSTROPHE	[']The straight single quote, doing duty as an apostrophe and as a string delimiter. ' is defined in XML and HTML5 but not in HTML 4, which is why ' is the safer escape.
40	28	050	00101000	(	LEFT PARENTHESIS
41	29	051	00101001	)	RIGHT PARENTHESIS
42	2A	052	00101010	*	ASTERISK
43	2B	053	00101011	+	PLUS SIGN
44	2C	054	00101100	,	COMMA
45	2D	055	00101101	-	HYPHEN-MINUS	One character doing the work of a hyphen, a minus sign and a dash. The typographic en dash, em dash and true minus all live outside ASCII.
46	2E	056	00101110	.	FULL STOP	Period, dot, point.
47	2F	057	00101111	/	SOLIDUS	The forward slash.
48	30	060	00110000	0	DIGIT ZERO	The digits run 48–57, so the numeric value of a digit character is its code minus 48.
49	31	061	00110001	1	DIGIT ONE
50	32	062	00110010	2	DIGIT TWO
51	33	063	00110011	3	DIGIT THREE
52	34	064	00110100	4	DIGIT FOUR
53	35	065	00110101	5	DIGIT FIVE
54	36	066	00110110	6	DIGIT SIX
55	37	067	00110111	7	DIGIT SEVEN
56	38	070	00111000	8	DIGIT EIGHT
57	39	071	00111001	9	DIGIT NINE
58	3A	072	00111010	:	COLON
59	3B	073	00111011	;	SEMICOLON
60	3C	074	00111100	<	LESS-THAN SIGN	[<]
61	3D	075	00111101	=	EQUALS SIGN
62	3E	076	00111110	>	GREATER-THAN SIGN	[>]
63	3F	077	00111111	?	QUESTION MARK
64	40	100	01000000	@	COMMERCIAL AT
65	41	101	01000001	A	LATIN CAPITAL LETTER A	Hex 41. The capitals occupy 65–90 with no gaps, and the lowercase run starts exactly 32 higher — so changing case is a single bit, which is why old code did it with an OR of 0x20.
66	42	102	01000010	B	LATIN CAPITAL LETTER B
67	43	103	01000011	C	LATIN CAPITAL LETTER C
68	44	104	01000100	D	LATIN CAPITAL LETTER D
69	45	105	01000101	E	LATIN CAPITAL LETTER E
70	46	106	01000110	F	LATIN CAPITAL LETTER F
71	47	107	01000111	G	LATIN CAPITAL LETTER G
72	48	110	01001000	H	LATIN CAPITAL LETTER H
73	49	111	01001001	I	LATIN CAPITAL LETTER I
74	4A	112	01001010	J	LATIN CAPITAL LETTER J
75	4B	113	01001011	K	LATIN CAPITAL LETTER K
76	4C	114	01001100	L	LATIN CAPITAL LETTER L
77	4D	115	01001101	M	LATIN CAPITAL LETTER M
78	4E	116	01001110	N	LATIN CAPITAL LETTER N
79	4F	117	01001111	O	LATIN CAPITAL LETTER O
80	50	120	01010000	P	LATIN CAPITAL LETTER P
81	51	121	01010001	Q	LATIN CAPITAL LETTER Q
82	52	122	01010010	R	LATIN CAPITAL LETTER R
83	53	123	01010011	S	LATIN CAPITAL LETTER S
84	54	124	01010100	T	LATIN CAPITAL LETTER T
85	55	125	01010101	U	LATIN CAPITAL LETTER U
86	56	126	01010110	V	LATIN CAPITAL LETTER V
87	57	127	01010111	W	LATIN CAPITAL LETTER W
88	58	130	01011000	X	LATIN CAPITAL LETTER X
89	59	131	01011001	Y	LATIN CAPITAL LETTER Y
90	5A	132	01011010	Z	LATIN CAPITAL LETTER Z
91	5B	133	01011011	[	LEFT SQUARE BRACKET
92	5C	134	01011100	\	REVERSE SOLIDUS	The backslash. It is the escape character in almost every language and in regular expressions, which is why writing one in a string means typing two.
93	5D	135	01011101	]	RIGHT SQUARE BRACKET
94	5E	136	01011110	^	CIRCUMFLEX ACCENT	The caret.
95	5F	137	01011111	_	LOW LINE	The underscore.
96	60	140	01100000	`	GRAVE ACCENT	The backtick.
97	61	141	01100001	a	LATIN SMALL LETTER A	Hex 61 — exactly 32 more than “A” at 41. Every letter keeps that offset, so an uppercase run and a lowercase run differ in one bit and nothing else.
98	62	142	01100010	b	LATIN SMALL LETTER B
99	63	143	01100011	c	LATIN SMALL LETTER C
100	64	144	01100100	d	LATIN SMALL LETTER D
101	65	145	01100101	e	LATIN SMALL LETTER E
102	66	146	01100110	f	LATIN SMALL LETTER F
103	67	147	01100111	g	LATIN SMALL LETTER G
104	68	150	01101000	h	LATIN SMALL LETTER H
105	69	151	01101001	i	LATIN SMALL LETTER I
106	6A	152	01101010	j	LATIN SMALL LETTER J
107	6B	153	01101011	k	LATIN SMALL LETTER K
108	6C	154	01101100	l	LATIN SMALL LETTER L
109	6D	155	01101101	m	LATIN SMALL LETTER M
110	6E	156	01101110	n	LATIN SMALL LETTER N
111	6F	157	01101111	o	LATIN SMALL LETTER O
112	70	160	01110000	p	LATIN SMALL LETTER P
113	71	161	01110001	q	LATIN SMALL LETTER Q
114	72	162	01110010	r	LATIN SMALL LETTER R
115	73	163	01110011	s	LATIN SMALL LETTER S
116	74	164	01110100	t	LATIN SMALL LETTER T
117	75	165	01110101	u	LATIN SMALL LETTER U
118	76	166	01110110	v	LATIN SMALL LETTER V
119	77	167	01110111	w	LATIN SMALL LETTER W
120	78	170	01111000	x	LATIN SMALL LETTER X
121	79	171	01111001	y	LATIN SMALL LETTER Y
122	7A	172	01111010	z	LATIN SMALL LETTER Z
123	7B	173	01111011	{	LEFT CURLY BRACKET
124	7C	174	01111100	\|	VERTICAL LINE	The pipe.
125	7D	175	01111101	}	RIGHT CURLY BRACKET
126	7E	176	01111110	~	TILDE
127	7F	177	01111111	␡	DEL · Delete	All seven bits set, so on paper tape it punched every hole in the column. That made it the one code you could over-punch any other character into, which is a physical delete you cannot undo. It sits at the end of the table rather than with the other controls because of that encoding, not by accident.still used

Control codes have no printable form, so the Char column shows a Unicode Control Picture — ␀, ␊, ␡ — which is a depiction of the code and not the code itself. Copying it will not give you the byte; use the escape shown in the description.

Above 127 — there is no “extended ASCII”

ASCII is a 7-bit code and ends at 127. Everything above that belongs to some other encoding, and those encodings disagree with each other. Below are fourteen byte values shown in three of the most common — ISO 8859-1 (Latin-1), Windows-1252, and IBM CP437, the original PC set. Same byte, three different characters. This is why “extended ASCII” names nothing you can act on, and why a file opened under the wrong assumption shows â€™ where an apostrophe belongs.

Dec	Hex	ISO 8859-1	Windows-1252	CP437
128	80	unassigned (C1)	€	Ç
133	85	unassigned (C1 NEL)	…	à
145	91	unassigned (C1)	‘	æ
146	92	unassigned (C1)	’	Æ
147	93	unassigned (C1)	“	ô
148	94	unassigned (C1)	”	ö
150	96	unassigned (C1)	–	û
151	97	unassigned (C1)	—	ù
153	99	unassigned (C1)	™	Ö
160	A0	no-break space	no-break space	á
176	B0	°	°	░
192	C0	À	À	└
233	E9	é	é	Θ
255	FF	ÿ	ÿ	no-break space

Only a sample, deliberately. Printing one of these encodings in full and labelling it “extended ASCII” would repeat the exact mistake this section exists to correct. If you need a complete high-range table, the encoding’s own specification is the source to use, and you have to know which encoding you have before you open it.

How it works

1
Look something up in either direction
The box at the top takes whichever thing you already have. Paste a character and it returns the decimal, hexadecimal, octal and binary values, the official name, the escape sequence and the HTML entity where one exists. Type 65, 0x41, \x41, U+0041, 0o101 or 0b1000001 and it returns “A”. There is no direction switch, because there is nothing ambiguous to switch between.
2
Read the control codes as what they were
Codes 0 to 31 and 127 are not characters, and describing them as “unprintable” explains nothing. They were commands to a teleprinter: BEL rang a real bell, CR returned the print head to column 1, LF wound the paper up one line, and BS moved back one position so the next letter struck over the last. Every one carries a line on the job it was built for and whether it still has one.
3
Filter when you know part of the answer
The filter matches the character, the name, the abbreviation and every numeric form, so “0x4” narrows to one run of the table and “separator” finds the four data separators in one go. It also has a control-codes-only view, which is the half of ASCII most tables print without explaining.
4
Check the arithmetic yourself
The table is built to be verified rather than trusted. The digits run 48 to 57, so a digit’s value is its code minus 48. Capitals run 65 to 90 and lowercase 97 to 122 — exactly 32 apart, which is one bit, which is why case conversion used to be a single OR with 0x20. None of that needs a source; you can read it off the rows.
5
Then read the correction at the bottom
Above 127 the table stops and the comparison starts. Byte 128 is undefined in Latin-1, a euro sign in Windows-1252 and a Ç in CP437, and that is not an edge case — it is the reason a file opened with the wrong assumption shows â€™ where an apostrophe should be.

Frequently asked questions

Why does ASCII stop at 127 instead of 255?

Because it is a 7-bit code, and 2^7 is 128. It was designed in the early 1960s for teleprinters and paper tape, where the eighth bit of a byte was often a parity bit used to detect a transmission error rather than to carry data. That is the whole answer: there was no eighth bit to spend. Every character above 127 that you have ever seen came from something built on top of ASCII afterwards, not from ASCII.

What is “extended ASCII”, then?

A phrase with no single meaning, which is why it causes so much trouble. Once the eighth bit became available, dozens of incompatible schemes filled 128–255 with different characters. ISO 8859-1 (Latin-1) put Western European accents there and left 128–159 unassigned. Windows-1252 filled that same gap with curly quotes, dashes and the euro sign. IBM’s CP437, the original PC character set, used the range for box-drawing characters and Greek letters instead. Byte 233 is “é” in the first two and “Θ” in CP437. Saying a file is “extended ASCII” tells you nothing about which of those it is — the comparison table on this page shows fourteen bytes where they disagree.

Why is DEL at 127, away from the other control codes?

Because of paper tape. 127 is 1111111 in binary — every bit set — so punching it made a hole in every position of the column. That meant any character already on the tape could be turned into DEL by punching the remaining holes, which is how you deleted something you could not un-punch. NUL sits at the other end for the mirror-image reason: all bits zero punched nothing at all, so it worked as blank leader tape and as filler.

Which control characters still matter today?

About a dozen. LF (10) is the newline on Unix and macOS and in every modern file format; CR (13) still pairs with it on Windows and is required by HTTP, SMTP and the other line-based protocols. HT (9) is the tab and the separator in TSV files. NUL (0) ends a C string. ESC (27) begins every ANSI terminal sequence, which is where terminal colour comes from. BEL (7) still beeps. Ctrl+C, Ctrl+D and Ctrl+Z are ETX, EOT and SUB. RS (30) is used by JSON Text Sequences and GS (29) by GS1 barcodes. The rest — SYN, ETB, DC2, EM — belong to synchronous links and paper tape and are effectively archaeology.

Is ASCII the same as UTF-8?

For the first 128 code points, byte for byte, yes — and that was deliberate. UTF-8 encodes anything below 128 as a single byte with the same value ASCII gives it, so every ASCII file is already a valid UTF-8 file. Above 127 they part company completely: UTF-8 uses two to four bytes per character, all with the high bit set. This is why ASCII text needs no conversion and why a Latin-1 file does — Latin-1’s “é” is one byte, 233, while UTF-8’s is two bytes, 195 and 169. Reading the second as the first is what produces “Ã©”.

Why do the control codes show a symbol instead of a character?

Because they have no printable form, and putting something in that column would be an invention. What is shown is a Unicode Control Picture — ␀ for NUL, ␊ for LF, ␡ for DEL — from the U+2400 block, which exists precisely so a table can depict a control code without pretending it is one. It is a picture of the character, not the character, and copying it will not give you the byte.

How do I type a character when I only have its code?

It depends on where you are typing. In JavaScript, String.fromCharCode(65) or the escape \u0041 inside a string. In HTML, A in decimal or A in hex — and for five characters you should use the named entity instead: & < > " and '. In a Unix shell, printf '\101' takes octal. On Linux many applications accept Ctrl+Shift+U followed by the hex code point. The lookup box above prints the escape for the character you are looking at, so you can copy the right form rather than remember it.

Does anything I type here get sent anywhere?

No. The whole table is part of the JavaScript this page already loaded, so the lookup is a map read in your browser and nothing else. There is no request, and nothing is written to localStorage, so the page works with the network off and closing the tab leaves nothing behind.

Where do hex, octal and binary come from if only decimal is stored?

They are formatted from the decimal value at display time, and that is deliberate. All four columns are the same integer written in four bases, so a conversion cannot disagree with the row it sits in — whereas typing 512 numbers by hand would add 384 chances to make a typo that no reader could catch. What is written out by hand is everything a machine cannot derive: the names, the abbreviations, the Ctrl keys, and what each control code was built to do.

How it works

Look something up in either direction

Read the control codes as what they were

Filter when you know part of the answer

Check the arithmetic yourself

Then read the correction at the bottom