The cut Command in Linux – A Complete Usage Guide

`records.txt` contains `17:Ada:engineer`, but `cut -c 2-4` takes character positions instead of returning the `Ada` field. I love that one compact command can give you a focused view of the row without changing the file.

That output can be passed along while the original record stays intact. `cut` makes its selection line by line, which is the key to understanding what it returns.

What cut selects from each line

The cut utility selects a portion of each input line by character position or delimiter, then writes that selection to standard output. It reads standard input when no file is named.

printf 'Linux shell\n' | cut -c 1-5

The command prints Linux because positions 1 through 5 are selected from the line. The same selection rule applies to every line in a file or stream.

Choose positions or fields before you run cut

The option sets what each number counts. Pick positions for fixed-width text and fields when one delimiter separates values.

Use one selection mode per command. The Linux manual pages show the options supported by the installed command.

Select characters and bytes by position

Positions start at 1, and both ends of a range are included. A comma selects separate positions, while a hyphen selects a continuous range.

cut -c 2-4 records.txt

A comma selects separate positions rather than a continuous range. In this command, positions 2, 5, and 7 are kept from each line.

cut -c 2,5,7 hello.txt

For the first line, name, the selected characters are ame, while -b selects byte positions when the list is measured in bytes.

Extract fields with a delimiter

Field mode uses a tab as its delimiter by default, and -d selects one other character for a file whose words have single spaces between them.

cut -d' ' -f1 hello.txt

When adjacent fields belong together, a range keeps the complete sequence in order. The command below selects fields 1 through 3.

cut -d " " -f 1-3 hello.txt

Without a filename, cut reads from standard input. The cat and more commands can combine files or show their contents before you select fields.

printf 'Ada\tadmin\nLin\teditor\n' | cut -f2

Keep other fields or change the output

GNU cut’s –complement option keeps unselected fields or positions. –output-delimiter changes the separator between selected fields, not the field contents.

• Use –complement to invert the selection.
• Use –output-delimiter to choose a new separator between fields.
• Use the shell redirection operator to save standard output to a file.

echo "Hello Journaldev" | cut -c -3 --complement

For selected fields, GNU cut can replace the input separator with a different output separator.

cut -d " " -f 1,2 hello.txt --output-delimiter='$'

Redirecting the same command writes its output to c.txt instead of the terminal. The file keeps the selected fields and the chosen separator.

cut -d " " -f 1,2 hello.txt --output-delimiter='$' > c.txt
cat c.txt

Handle missing delimiters and uneven whitespace

In field mode, a line with no delimiter passes through unchanged unless you add -s to suppress it.

cut -d: -f2 records.txt

With -s, only lines containing the delimiter remain. The screenshot shows the input row that disappears from the result.

cat records.txt && cut -s -d: -f2 records.txt

Because each space is its own delimiter, repeated spaces create empty fields and shift later values.

If you process several files, a Bash for loop can apply the same command to each file. A Bash while loop fits a task where each line needs its own decision, and the line-count command can check how many records remain after filtering.

Know when cut is the wrong parser

cut treats delimiters as literal characters, not as CSV quoting rules or a run of whitespace. In the example below, the comma inside the quoted value still separates fields.

printf 'name,description\nAda,"blue, green"\n' | cut -d, -f2

The second output line is “blue, not the complete quoted value. A CSV-aware parser is the right choice when quoted separators belong inside a field.

The POSIX specification defines -c as character selection and -b as byte selection. I found that position 4 returned byte C3 from café in the GNU Coreutils 9.4 test under C.UTF-8, followed by 0A for the newline. So check the installed implementation before using -c to slice multibyte text.

printf 'café\n' | cut -c 4 | od -An -tx1

For comparison, awk can count characters in a UTF-8 locale, while a CSV parser can honor quoted commas. Choose the parser from the data format, not from the shape of one sample row.

printf 'café\n' | LC_ALL=C.UTF-8 awk '{print substr($0,4,1)}'

Keep cut at the line boundary

cut works well when each line has a stable delimiter or fixed positions and the same extraction applies to every row. Once quoting, variable whitespace, or row-specific logic changes the result, switch to a parser that understands that structure.

printf 'id:name:role\n17:Ada:engineer\n24:Lin:admin\nNoDelimiter\n' | cut -s -d: -f2 | sort -u

Frequently asked questions

The answers below cover the delimiter and input questions that change how cut behaves.