Viewing files

`od`

Dump a file's bytes as octal, hex, decimal or characters, to see exactly what it contains.

od [OPTION...] [FILE...]

9 of 9 examples run in Debian 13.7 on 2026-10-31 · coreutils 9.7-3 · manual: od(1)

od (octal dump) prints the raw contents of a file in a readable notation. Its default output is 2-byte octal words, which is rarely what anyone wants today; the useful forms are od -c (each byte as a character or an escape like \r or \n) and od -t x1 (each byte in hex).

od is part of coreutils and POSIX, so it is on every Unix system, which makes it the safe choice in scripts. For interactive hex viewing, xxd and hexdump -C print a friendlier side-by-side hex and text layout. Use od when you need its -t type control: decoding bytes as 4-byte integers, floats, or with a chosen byte order.

Each output line starts with the byte offset, in octal unless you change it with -A.

Examples

See every byte as a character

$ od -c data/windows.txt

explain

output

0000000   l   i   n   e       o   n   e  \r  \n   l   i   n   e       t
0000020   w   o  \r  \n
0000024

-c shows printable bytes as themselves and control bytes as C escapes, so the \r \n pairs reveal CRLF line endings. The offsets on the left are octal: 0000020 is byte 16.

Hex bytes without offsets

$ od -An -t x1 todo.txt | head -n 2

explain

output

 62 75 79 20 6d 69 6c 6b 0a 66 69 78 20 6c 6f 67
 69 6e 20 62 75 67 0a 77 72 69 74 65 20 74 65 73

-t x1 prints one byte per hex number and -An drops the offset column. 62 75 79 is buy, 20 a space and 0a the newline.

Hex next to printable text

$ od -A d -t x1z -N 32 photos/beach.png

explain

output

0000000 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 52  >.PNG........IHDR<
0000016 00 00 02 80 00 00 01 e0 08 02 00 00 00 ba b3 4b  >...............K<
0000032

The z suffix adds the printable characters between > and < at the end of each line, -A d makes the offsets decimal and -N 32 stops after 32 bytes. The .PNG signature and the IHDR chunk name stand out.

Read a few bytes from a given offset

$ od -A d -t x1 -j 16 -N 8 photos/beach.png

explain

output

0000016 00 00 02 80 00 00 01 e0
0000024

-j 16 skips the first 16 bytes and -N 8 reads 8. In a PNG those are the image width and height: 00 00 02 80 and 00 00 01 e0.

Decode bytes as big-endian integers

$ od --endian=big -An -t u4 -j 16 -N 8 photos/beach.png

explain

output

        640        480

-t u4 reads 4-byte unsigned integers. PNG stores numbers big-endian, so --endian=big decodes them correctly: the image is 640 by 480.

The machine's byte order by default

$ printf '\001\000\000\000' | od -An -t d4

explain

output

Without --endian, multi-byte values use the CPU's byte order. On this little-endian x86 machine, the bytes 01 00 00 00 are the number 1; a big-endian reading would give 16777216.

Byte values in decimal

$ printf '\377\376A' | od -An -t u1

explain

output

 255 254  65

-t u1 prints each byte as an unsigned decimal number: 255, 254 and 65 (the letter A).

Change how many bytes per line

$ od -An -c -w8 todo.txt | head -n 3

explain

output

   b   u   y       m   i   l   k
  \n   f   i   x       l   o   g
   i   n       b   u   g  \n   w

-w8 puts 8 bytes on each line instead of 16, which is easier to read for short records.

The historical default format

$ od README.md | head -n 2

explain

output

0000000 020043 062544 067555 005012 020101 066563 066141 020154
0000020 060563 070155 062554 070040 067562 062552 072143 072440

Plain od prints 2-byte words in octal, in machine byte order, which is hard to relate to the text. This is why nearly everyone types od -c or od -t x1.

Common pitfalls

Offsets are octal by default: 0000020 means byte 16, not 20. Add -A d for decimal or -A x for hex offsets.
Multi-byte types (-t x2, -t d4, plain od) use the CPU's byte order, so the same file dumps differently on different machines. Use -t x1 to see bytes as they are stored, or --endian.
Repeated identical lines are collapsed to a single *. Add -v when you need every line, for example to diff two dumps.

GNU vs BSD and macOS

The examples above ran the GNU/Linux version. These are the differences that matter on macOS and the BSDs (written from their documentation, not run here).

--endian, -w and the z suffix on -t are GNU additions. For scripts that must also run on BSD/macOS, stick to the POSIX options -A, -j, -N, -t, -c and -v.

Options worth knowing

A selection, described in our own words. For the full list, see od(1) or od --help.

Option	What it does
`-c`	Show bytes as printable characters or C-style escapes; same as -t c.
`-t, --format TYPE`	Output type: c, a, x1, x2, o1, u1, d4, f8 and so on; add z to append printable text.
`-A, --address-radix RADIX`	Offset format: d (decimal), o (octal, default), x (hex) or n (none).
`-j, --skip-bytes BYTES`	Skip this many bytes of input first.
`-N, --read-bytes BYTES`	Dump at most this many bytes.
`-v, --output-duplicates`	Print every line; do not replace repeats with *.
`-w, --width[=BYTES]`	Bytes per output line (32 if given without a value).
`--endian ORDER`	Byte order for multi-byte types: big or little.
`-x`	Same as -t x2: 2-byte hex words.
`-b`	Same as -t o1: 1-byte octal.
`-a`	Same as -t a: named characters, ignoring the high bit.

Examples

Common pitfalls

GNU vs BSD and macOS

Options worth knowing

Related