Bash remove from array: one element, first item, duplicates, and subtract another array

bash remove element from array and bash remove item from array usually mean either “drop this value”, “drop the first/last slot”, bash array remove duplicates, or “everything in A that is not listed in B” (the original title of this page). Pick one job; mixing them is how scripts grow grep pipes that break on spaces. If your data started as a string, split first with bash split string into array.

Snippets below were run with Bash 5.2.37 on Ubuntu 25.04 (kernel 6.14.0-37-generic).

bash remove item from array (filter by value)

Build a new array and skip the value you do not want—safe with spaces and glob characters because the array stays quoted:

array1=(a b c d)
remove=b
filtered=()
for item in "${array1[@]}"; do
  [[ $item == "$remove" ]] && continue
  filtered+=("$item")
done
printf '%s ' "${filtered[@]}"
echo

a c d

That is the usual answer to bash remove from array / remove element from array bash when you only remove one known token.

bash remove first element from array (or tail by slice)

Slices are the fast path for “drop head or tail” without scanning:

nums=(10 20 30 40)
rest=("${nums[@]:1}")     # drop first
echo "${rest[@]}"

20 30 40

Drop the last element on Bash 4.3+:

nums=(10 20 30)
unset 'nums[-1]'
echo "${nums[@]}"

10 20

bash array remove duplicates (uniq, or order-preserving map)

Sorted input, order not important:

vals=(b a b c a)
printf '%s\n' "${vals[@]}" | sort -u

a
b
c

bash remove duplicates from array while keeping first-seen order needs an associative array (Bash 4+):

vals=(a b a c b)
declare -A seen=()
deduped=()
for v in "${vals[@]}"; do
  [[ -n ${seen[$v]+isset} ]] && continue
  seen[$v]=1
  deduped+=("$v")
done
printf '%s ' "${deduped[@]}"
echo

a b c

Delete elements of one array from another (set difference A \ B)

Mark everything in B, walk A, skip hits—O(n) and robust for arbitrary strings (unlike ${a[@]/x/} tricks that leave empty slots or partial substring matches):

A=(a b c d)
B=(b e)
declare -A drop=()
for x in "${B[@]}"; do drop[$x]=1; done
out=()
for x in "${A[@]}"; do
  [[ -n ${drop[$x]+isset} ]] && continue
  out+=("$x")
done
printf '%s ' "${out[@]}"
echo

a c d

That is the clean bash remove item from list pattern when the “list” to remove lives in a second array.

Why not ${array[@]/pattern/} for removal?

${arr[@]/b/} substitutes substring matches inside every element and can leave empty entries ("") instead of removing a slot. It also treats patterns as substrings, not “whole cell equals b”. Prefer explicit loops or an associative map.

comm / grep shortcuts (when data is simple)

If every token is a single word with no spaces, sorted comm -23 can print A \ B:

A=(a b c d)
B=(b e)
comm -23 <(printf '%s\n' "${A[@]}" | sort) <(printf '%s\n' "${B[@]}" | sort) | paste -sd' ' -

a c d

Keep the map version when elements can contain spaces or you care about original order.

Related

• bash concatenate strings when building lines from arrays.
• beginner Bash arguments for passing array-like data into scripts.
• prepend to arrays and positional parameters (unshift patterns in Bash) when you are reshaping lists at the head as well as the tail.

Summary

Use a quoted loop for bash remove element from array by value, "${arr[@]:1}" or unset 'arr[-1]' for bash remove first element from array (or trimming the tail), sort -u or an declare -A pass for bash array remove duplicates / bash remove duplicates from array, and an associative “drop set” for delete elements of one array from another. Avoid ${array[@]/x/} unless you truly want substring substitution with possible empty cells; prefer explicit tests for bash remove item from list clarity.