A one-liner that removes a column from a CSV stops on a Mac:

$ mkdir demo && cd demo
$ printf 'id,name,email,city\n1,ann,a@example.com,Oslo\n2,bob,b@example.com,Lima\n' > users.csv
$ cut --complement -d, -f3 users.csv; echo "exit $?"
cut: illegal option -- -
usage: cut -b list [-n] [file ...]
       cut -c list [file ...]
       cut -f list [-s] [-w | -d delim] [file ...]
exit 1

Moving --complement after -d, gives the same error:

$ cut -d, --complement -f3 users.csv; echo "exit $?"
cut: illegal option -- -
usage: cut -b list [-n] [file ...]
       cut -c list [file ...]
       cut -f list [-s] [-w | -d delim] [file ...]
exit 1

--complement is a GNU cut option (I didn’t run GNU cut here). macOS ships BSD cut; its usage text above lists no long options, and the man page doesn’t mention a complement:

$ what /usr/bin/cut
/usr/bin/cut:
	PROGRAM:cut  PROJECT:text_cmds-199
	PROGRAM:cut  PROJECT:text_cmds-199
$ man cut | col -b | grep -ci complement
0

Fix 1: name the fields you keep, with an open range

cut -f takes a list, and a range may be open on the right. To drop field 3, keep 1–2 and 4-to-the-end:

$ cut -d, -f1-2,4- users.csv
id,name,city
1,ann,Oslo
2,bob,Lima

4- means “field 4 and everything after”, so this doesn’t need the field count. The same holds on lines with more or fewer fields:

$ printf 'a,b,c,d,e,f\nx,y\nnodelim\n' > ragged.csv
$ cut -d, -f1-2,4- ragged.csv
a,b,d,e,f
x,y
nodelim

The last line has no comma, so cut printed it unchanged. That’s documented: “Unless specified, lines with no delimiters are passed through unmodified.” Add -s to drop such lines instead:

$ cut -s -d, -f1-2,4- ragged.csv
a,b,d,e,f
x,y

The list doesn’t reorder fields. -f4,1 still prints them in input order:

$ cut -d, -f4,1 users.csv
id,city
1,Oslo
2,Lima

The -f1-0 trap when dropping field 1

If you build the list as 1-(N-1),(N+1)-, N=1 gives 1-0, and cut rejects it:

$ cut -d, -f1-0,2- users.csv; echo "exit $?"
cut: [-bcf] list: values may not include zero
exit 1
$ cut -d, -f-0,2- users.csv; echo "exit $?"
cut: [-bcf] list: values may not include zero
exit 1
$ cut -d, -f2- users.csv
name,email,city
ann,a@example.com,Oslo
bob,b@example.com,Lima

A small function that handles both cases:

cut_drop() {  # usage: cut_drop FIELD DELIM [file ...]
  n=$1 d=$2; shift 2
  if [ "$n" -eq 1 ]; then
    cut -d "$d" -f 2- "$@"
  else
    cut -d "$d" -f "1-$((n - 1)),$((n + 1))-" "$@"
  fi
}

$ cut_drop 1 , users.csv
name,email,city
ann,a@example.com,Oslo
bob,b@example.com,Lima
$ cut_drop 3 , users.csv
id,name,city
1,ann,Oslo
2,bob,Lima
$ cut_drop 1 , ragged.csv
b,c,d,e,f
y
nodelim

The nodelim line is still passed through — even when you “drop” its only field.

Fix 2: awk, and two awk shortcuts that go wrong

Setting the field to empty does not remove it. The separator stays:

$ awk -F, -v OFS=, '{ $3 = "" } 1' users.csv
id,name,,city
1,ann,,Oslo
2,bob,,Lima

Build the line from the fields you keep instead:

$ awk -F, -v OFS=, -v c=3 '{ o = ""; s = ""; for (i = 1; i <= NF; i++) if (i != c) { o = o s $i; s = OFS }; print o }' users.csv
id,name,city
1,ann,Oslo
2,bob,Lima
$ awk -F, -v OFS=, -v c=3 '{ o = ""; s = ""; for (i = 1; i <= NF; i++) if (i != c) { o = o s $i; s = OFS }; print o }' ragged.csv
a,b,d,e,f
x,y
nodelim

The s variable (empty before the first kept field, OFS after) keeps a leading empty field intact. A version that checks o == "" to decide whether to add a separator loses it:

$ printf ',b,c\n' | awk -F, -v OFS=, -v c=3 '{ o = ""; s = ""; for (i = 1; i <= NF; i++) if (i != c) { o = o s $i; s = OFS }; print o }'
,b
$ printf ',b,c\n' | awk -F, -v OFS=, -v c=3 '{ o = ""; for (i = 1; i <= NF; i++) if (i != c) o = (o == "" ? $i : o OFS $i); print o }'
b

Unlike cut, awk doesn’t pass nodelim through when you drop field 1 — the line has one field, you removed it, and an empty line is left:

$ awk -F, -v OFS=, -v c=1 '{ o = ""; s = ""; for (i = 1; i <= NF; i++) if (i != c) { o = o s $i; s = OFS }; print o }' ragged.csv
b,c,d,e,f
y

A shorter version you’ll see around shifts the fields left and decrements NF. In this awk it works on full lines but not on short ones:

$ awk -F, -v OFS=, -v c=3 '{ for (i = c; i < NF; i++) $i = $(i + 1); NF--; print }' ragged.csv
a,b,d,e,f
x

x,y has no field 3, but NF-- still removed the last field (y), and nodelim became an empty line.

Delimiters: where cut and awk differ

Default is tab, not whitespace. Without -d, cut splits on tab. A CSV has no tabs, so every line is “without delimiter” and comes back whole, with exit 0:

$ cut -f2 users.csv; echo "exit $?"
id,name,email,city
1,ann,a@example.com,Oslo
2,bob,b@example.com,Lima
exit 0

-d ' ' counts every space. Two spaces in a row make an empty field. awk’s default field splitting doesn’t:

$ printf 'web  8080 up\ndb 5432 down\n' > ps.txt
$ cut -d ' ' -f2 ps.txt

5432
$ awk '{ print $2 }' ps.txt
8080
5432

BSD cut has -w. From the man page: “Use whitespace (spaces and tabs) as the delimiter. Consecutive spaces and tabs count as one single field separator.” The output fields are joined with a tab:

$ cut -w -f2 ps.txt
8080
5432
$ cut -w -f1,3- ps.txt | od -c
0000000    w   e   b  \t   u   p  \n   d   b  \t   d   o   w   n  \n
0000017

awk’s default rebuild uses single spaces. Once you assign to a field with the default FS, awk rejoins the line with OFS (a space), so the original spacing is gone, and the emptied field leaves two spaces:

$ awk '{ $2 = "" } 1' ps.txt | od -c
0000000    w   e   b           u   p  \n   d   b           d   o   w   n
0000020   \n
0000021

One character only. cut won’t take a multi-character delimiter:

$ cut -d ', ' -f1 users.csv; echo "exit $?"
cut: bad delimiter
exit 1

Quoted CSV is not handled by either. A comma inside quotes is still a separator to both:

$ printf '1,"Lee, Ann",x@example.org\n' | cut -d, -f1-2,4-
1,"Lee,x@example.org
$ printf '1,"Lee, Ann",x@example.org\n' | awk -F, '{ print NF }'
4

For real CSV with quoted fields, use a CSV parser (for example Python’s csv module) — not tested in this post.

Checklist

  • cut: illegal option -- - (exit 1 here): /usr/bin/cut on macOS is BSD cut; its usage text lists no long options and its man page has no complement.
  • Drop field N: keep 1-(N-1),(N+1)-. The open - end means you don’t need the field count.
  • Dropping field 1: use -f2-. -f1-0 and -f-0,2- fail with “values may not include zero”.
  • Lines with no delimiter come through unchanged; -s drops them.
  • -f4,1 prints in input order, not 4 then 1.
  • awk: $3 = "" leaves an empty column. Rebuild from kept fields; avoid NF-- shortcuts on lines that may be short.
  • cut default delimiter is tab; -d ' ' counts every space; -w treats runs of whitespace as one and outputs tabs; a two-character -d is rejected.
  • Neither cut nor awk -F, understands quoted CSV.

Tested on macOS 26.5.2 (Apple silicon) with /usr/bin/cut (PROGRAM:cut PROJECT:text_cmds-199) and /usr/bin/awk (version 20200816) in zsh 5.9 (arm64-apple-darwin25.0). GNU cut, gawk, bash, and Linux were not tested; cut -b/-c were not tested.