sed Delete Lines: Remove Lines by Number, Pattern, or Range

sed processes input line by line, applies your commands, and writes the result to standard output. The original file stays untouched unless you pass the -i flag. This makes it safe to preview any operation before committing it.

The general form of the delete command is:

bashsed 'ADDRESSd' file

ADDRESS selects which lines to delete. d is the delete command. Lines that do not match the address print as usual.

How to Use the sed Delete Command

Delete a specific line by number:

bashsed '3d' file.txt     # delete line 3sed '$d' file.txt     # delete the last line

$ always refers to the last line regardless of file length, useful when you need to remove a trailing entry without counting lines first.

Delete multiple non-consecutive lines by separating addresses with a semicolon:

bashsed '2d;5d;8d' file.txt

Delete Lines by Range, Invert, and Every Nth Line

Delete a block of consecutive lines with a range:

bashsed '3,6d' file.txt    # delete lines 3 through 6sed '5,$d' file.txt    # delete from line 5 to end of file

Adding ! after an address inverts the selection — it deletes everything that does NOT match. To keep only lines 3 through 5 and remove the rest:

bashsed '3,5!d' file.txt

This is a compact alternative to chaining head and tail to extract a slice.

GNU sed supports the first~step step address, which matches every Nth line starting at a given offset. To delete every even-numbered line:

bashsed '0~2d' file.txt

The offset 0 means the first match falls on line 2 (before the file starts, step of 2, so first match = line 2). To delete every third line starting from line 3:

bashsed '3~3d' file.txt

The first~step syntax is GNU sed only. It is not available in the BSD sed shipped with macOS.

Delete Lines Matching a Pattern

Wrap a search pattern in slashes to delete matching lines:

bashsed '/error/d' file.txt

Pattern matching is a substring match — /bl/d would delete any line containing “bl”, including “black”. Be precise with your patterns.

To match case-insensitively, add the I flag after the closing slash:

bashsed '/Error/Id' file.txt

To invert the match and delete all lines that do NOT contain the pattern:

bashsed '/error/!d' file.txt

Use two patterns separated by a comma to delete a range between matching lines:

bashsed '/start/,/end/d' file.txt

sed starts deleting at the first line matching the opening pattern and stops after the first match of the closing pattern. Both endpoints are included in the deletion.

Mix a line number with a pattern when you know where the range starts but not where it ends:

bashsed '1,/header_end/d' file.txt

Delete Empty Lines and Use Regex

Remove truly empty lines with ^$ — the beginning anchor immediately followed by the end anchor:

bashsed '/^$/d' file.txt

This does not catch lines that contain only spaces or tabs. To remove those as well, use the POSIX [[:space:]] character class:

bashsed '/^[[:space:]]*$/d' file.txt

* matches zero or more whitespace characters, so this covers both empty lines and visually blank lines.

The pattern between slashes is a full regular expression:

bashsed '/^#/d' config.txt           # strip comment lines starting with #sed '/;$/d' source.txt           # remove lines ending with semicolonsed -E '/^.{0,4}$/d' file.txt   # delete lines shorter than 5 characters

-E enables extended regular expressions — no need to escape { and }.

You can chain multiple commands in one expression. Commands execute left to right, and deleted lines never reach later ones:

bashsed '/^#/d; s/localhost/127.0.0.1/g' config.txt

This strips comment lines first, then applies the substitution only to remaining lines.

Edit Files In Place with -i

All the examples above write to standard output. To modify the file permanently, add -i:

bashsed -i '/^$/d' file.txt

Always preview without -i first. When ready, pass a suffix to save an automatic backup:

bashsed -i.bak '/^$/d' file.txt

This saves the original as file.txt.bak before modifying file.txt.

Never redirect output to the same input file — the shell truncates it before sed reads it:

bashsed '/^$/d' file.txt > file.txt    # destroys the file

On macOS, BSD sed requires an explicit extension argument even when you do not want a backup file:

bashsed -i '' '/^$/d' file.txt

Use line numbers for exact positions, ! to invert selection, patterns for content-based deletion, and [[:space:]] for blank line cleanup. Always preview without -i first, and use -i.bak when editing in place. Leave a comment below if you run into any issues.