sort — quick reference
Basic sorting
Read lines from files or stdin and print them in order. With no file argument, sort reads standard input.
| When to use | Command |
|---|---|
| Sort lines alphabetically (default locale order) | sort file.txt |
| Sort lines from a pipe | command | sort |
| Reverse the sort order | sort -r file.txt |
| Fold upper and lower case together | sort -f file.txt |
| Stable sort — keep original order among equal keys | sort -s file.txt |
| Output only the first line in each run of equal keys | sort -u file.txt |
Keys, fields, and delimiters
Sort by columns instead of the whole line. Fields default to whitespace-separated tokens unless you set -t.
| When to use | Command |
|---|---|
| Sort by the second field | sort -k2 file.txt |
| Sort by field 2 numerically | sort -k2,2n file.txt |
| Sort CSV by the second column (comma delimiter) | sort -t',' -k2,2n file.csv |
| Ignore leading blanks when finding sort keys | sort -b file.txt |
| Consider only letters, digits, and blanks | sort -d file.txt |
| Ignore non-printing characters | sort -i file.txt |
Numeric and special sort modes
Pick a comparison that matches the data — plain numbers, human sizes, months, or version strings.
| When to use | Command |
|---|---|
Sort by numeric value (10 after 2) |
sort -n file.txt |
| Sort general numeric values (scientific notation) | sort -g file.txt |
Sort human-readable sizes (1M > 100K) |
sort -h file.txt |
Sort version strings (1.12.2 > 1.1.2) |
sort -V file.txt |
| Sort by three-letter month names | sort -M file.txt |
| Shuffle lines randomly | sort -R file.txt |
Files, output, and merging
Write results to a file, merge already-sorted inputs, or handle NUL-terminated records.
| When to use | Command |
|---|---|
| Write sorted output to a file (safe in-place: same path for input and output) | sort -o file.txt file.txt |
| Merge pre-sorted files without resorting | sort -m sorted1.txt sorted2.txt |
| Use NUL instead of newline as the line delimiter | sort -z file.list |
| Read input file list from a NUL-separated file | sort --files0-from=list.nul |
Checking sorted input
Verify order without rewriting the file — useful in scripts and CI.
| When to use | Command |
|---|---|
| Report the first out-of-order line | sort -c file.txt |
| Exit 0 only if already sorted (silent) | sort -C file.txt |
Help and version
| When to use | Command |
|---|---|
| Show built-in usage | sort --help |
| Show implementation version | sort --version |
sort — command syntax
Synopsis from sort --help on Ubuntu 25.04 (uutils coreutils 0.2.2):
sort [OPTION]... [FILE]...Key format: FIELD[.CHAR][OPTIONS][,FIELD[.CHAR]][OPTIONS] — fields start at 1. Per-key option letters (n, r, h, V, …) override global flags for that key only. sort does not modify system files unless you pass -o to overwrite an output path.
sort — command examples
Essential Sort lines alphabetically
The default mode orders lines by locale rules — the starting point for any text file or command output you want in reading order.
Run the command:
printf '%s\n' zebra Apple banana cherry Banana | sortSample output:
Apple
Banana
banana
cherry
zebraUppercase letters sort before lowercase in the default C/POSIX locale. Add -f when case should not matter.
Essential Sort numbers numerically
Without -n, sort compares digit strings character by character (10 before 2). Use -n for integer-like columns.
Run the command:
printf '%s\n' 5108 152 126 16 529 70 1 990 56 | sort -nSample output:
1
16
56
70
126
152
529
990
5108Pipe ls, du, or awk output through sort -n when the column holds counts or byte sizes.
Common Reverse order and ignore case
Combine -r for descending order and -f to treat Apple and apple as the same letter.
Run the command:
printf '%s\n' zebra Apple banana cherry Banana | sort -rfSample output:
zebra
cherry
banana
Banana
AppleUse this on mixed-case hostnames or usernames when you want a single alphabetical list.
Common Sort a CSV file by the second column
-t sets the field separator; -k2,2n sorts only field two as a number.
Run the command:
printf '%s\n' 'bob,30' 'alice,25' 'carol,30' | sort -t',' -k2,2nSample output:
alice,25
bob,30
carol,30When two rows share the same key (carol and bob both have 30), their relative order depends on the rest of the line unless you add -s.
Common Sort human-readable sizes from du or ls
-h understands suffixes like K, M, and G so 2M sorts above 100K.
Pipe du output through sort -h for largest folders first; see the df and du.
Run the command:
printf '%s\n' 100K 2M 50k 1G | sort -hSample output:
50k
100K
2M
1GPair with du -h or custom awk output when building disk-usage reports.
Common Sort and drop duplicate lines
-u keeps one line per unique value after sorting. For stricter dedup of unsorted input, pipe to uniq instead.
Run the command:
printf '%s\n' dup alpha dup beta dup | sort -uSample output:
alpha
beta
dupEquivalent pipeline: sort file.txt | uniq > output.txt.
Common Sort a file in place with -o
-o writes to a named file. Using the same path for input and output replaces the file with sorted lines.
Run the command:
printf '%s\n' line3 line1 line2 > unsorted.txt
sort -o unsorted.txt unsorted.txt
cat unsorted.txtSample output:
line1
line2
line3Avoid sort file.txt > file.txt — the shell truncates the file before sort reads it.
Advanced Sort version strings with -V
Version sort treats dotted release numbers sensibly — 1.12.2 sorts after 1.1.2, unlike plain lexical order.
Run the command:
printf '%s\n' 1.12.2 1.1.2 1.2.0 | sort -VSample output:
1.1.2
1.2.0
1.12.2Use on package lists, kernel build strings, or semver tags in shell scripts.
Advanced Verify a file is already sorted
-c prints the first disorder and exits non-zero. -C is silent — handy in scripts that only need the exit code.
Run the command:
printf '%s\n' b a c > bad.txt
sort -c bad.txt 2>&1
echo exit:$?Sample output:
sort: bad.txt:2: disorder: a
exit:1Run sort -C good.txt in CI when a pipeline step must prove log lines arrived in order.
Advanced Merge two already-sorted files
-m combines ordered inputs in linear time without a full resort — useful after splitting huge sorted logs.
Run the command:
printf '%s\n' a c e > m1.txt
printf '%s\n' b d f > m2.txt
sort -m m1.txt m2.txtSample output:
a
b
c
d
e
fEach input file must already be sorted on the same key you care about.
sort — when to use / when not
| Use sort when | Use something else when |
|---|---|
|
sort vs ls sorting
ls can sort its own listing (ls -lS by size, ls -lt by time), but sort is the general text tool for any columnar or line-oriented data — file listings, logs, CSV, /etc/passwd fields, or find output. Piping ls -l to sort -k5 -n still works when you need a custom key or stable ordering ls does not offer.
Related commands
Text and file workflows that usually appear beside sort in scripts.
| Command | One line |
|---|---|
| sort | Order lines and fields (this page) |
| uniq | Drop adjacent duplicate lines (run after sort) |
| awk | Extract or reshape columns before sorting |
| cut | Pull one field, then pipe to sort |
Browse the full index in our Linux commands reference.
sort — interview corner
What does the sort command do in Linux?
sort reads lines from files or standard input and prints them in sorted order. By default it uses locale collation (alphabetical). Flags change the comparison: -n for numbers, -k for key fields, -r for reverse.
Example:
printf '%s\n' c b a | sortSample output:
a
b
cA strong answer is:
"sort orders text lines from files or stdin. I use -n for numeric columns, -k for fields, and -o to write results back to a file safely."
How does sort -k work?
-k selects which field (column) drives the comparison. Fields are numbered from 1 and separated by whitespace unless you set -t.
Syntax: sort -k START[,END][OPTIONS]. Example — sort by the second field numerically:
sort -k2,2n data.txtFor ls -l style output, sort -k5 -n sorts by the size column (fifth field).
A strong answer is:
"-k picks the sort key — field number from 1, with optional end field and n for numeric. I combine -t for CSV delimiters and -k2,2n when only one column should drive the order."
Why does sort put 10 before 2 without -n?
Default lexicographic sort compares characters left to right. The first character of 10 is 1, which comes before 2, so 10 appears before 2.
With -n:
printf '%s\n' 10 2 1 | sort -nSample output:
1
2
10A strong answer is:
"Without -n, sort compares strings character by character. -n parses fields as numbers so 2 comes before 10."
How do you sort a file in place?
Use -o to name the output file. When input and output are the same path, sort writes to a temp file first:
sort -o file.txt file.txtDo not use sort file.txt > file.txt — the shell empties the file before sort reads it.
A strong answer is:
"sort -o file.txt file.txt — the -o form is safe for in-place sorts. Redirection to the same path truncates the input."
What is the relationship between sort and uniq?
uniq only removes adjacent duplicate lines. Random duplicates in a file survive uniq until you sort first:
sort file.txt | uniq > unique.txtOr use sort -u, which sorts and keeps one line per unique value in a single step.
A strong answer is:
"uniq needs duplicates on consecutive lines, so I sort first. sort -u combines both steps when I only need unique sorted lines."
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
10 before 2 in output |
Lexicographic default | Add -n or -g for numeric data |
Empty file after sort f > f |
Shell truncation | Use sort -o f f |
| CSV columns mis-sorted | Wrong delimiter | sort -t',' -k2,2n file.csv |
sort: disorder with -c |
File not in expected order | Sort with the same keys, or fix upstream |
Unknown option on Ubuntu 25.04 |
uutils vs GNU flag drift | Run sort --help on the host; this page matches uutils coreutils 0.2.2 |
| Case order surprises | Locale collation | LC_ALL=C sort file for byte order, or -f to fold case |
| Version strings wrong order | Lexical sort | Use -V for dotted version numbers |
References
- sort(1) on man7.org — GNU coreutils reference on other distros
