PHP preg_replace(): Regex Replace with Backreferences

preg_replace($pattern, $replacement, $subject) finds every match of a regular expression in a string and replaces it. preg_replace('/\s+/', ' ', $text) turns every run of spaces, tabs and newlines into a single space.

<?php
$text = "Too    many \t spaces\n\nand   lines";
echo preg_replace('/\s+/', ' ', $text), "\n";

// Remove every digit
echo preg_replace('/\d/', '', 'R2-D2 and C-3PO'), "\n";

// Case-insensitive, whole words only
echo preg_replace('/\bcat\b/i', 'dog', 'Cat, cats and a CAT.'), "\n";

The original string is not changed; preg_replace() returns a new one. \b marks a word boundary, which is why cats is left alone.

Backreferences: $1,${1} and \1

In the replacement, $1 stands for whatever the first capture group matched, $2 for the second, and $0 for the whole match. Use them to rearrange the matched text.

<?php
// 2026-03-14 to 14/03/2026
echo preg_replace('/(\d{4})-(\d{2})-(\d{2})/', '$3/$2/$1', 'Due: 2026-03-14'), "\n";

// "Lovelace, Ada" to "Ada Lovelace"
echo preg_replace('/^(\w+), (\w+)$/', '$2 $1', 'Lovelace, Ada'), "\n";

// $0 is the whole match: wrap every number in brackets
echo preg_replace('/\d+/', '[$0]', 'Rooms 12 and 305'), "\n";

// ${1} when a digit follows the reference
echo preg_replace('/(\d+)/', '${1}0', 'Multiply 5 and 7 by ten'), "\n";

Write the replacement in single quotes. In double quotes "\1" becomes an octal escape (the byte 0x01) and "${1}" is read as a variable named 1, which is undefined and uses string syntax deprecated since PHP 8.2. '\1' in single quotes works the same as '$1', but $1 is the common spelling.

Remove non-alphanumeric characters

A negated character class, [^...], matches everything that is not listed. Replacing it with '' removes it. For text in other languages, use the Unicode classes \p{L} (any letter) and \p{N} (any digit) with the u modifier, or you will strip é, ü and every Japanese character.

<?php
$s = 'Héllo, wörld! 日本語 #2026';

echo preg_replace('/[^a-zA-Z0-9]/', '', $s), "\n";           // ASCII only
echo preg_replace('/[^a-zA-Z0-9 ]/', '', $s), "\n";          // keep spaces
echo preg_replace('/[^\p{L}\p{N}]/u', '', $s), "\n";         // any language
echo preg_replace('/[^\p{L}\p{N}\s]/u', '', $s), "\n";       // and spaces

// Digits only, for a phone number typed with spaces and dashes
echo preg_replace('/\D/', '', '+1 (555) 123-4567'), "\n";

Make a URL slug

Several patterns in a row turn a title into a URL slug. Lowercase first, replace every run of non-alphanumeric characters with one hyphen, then trim hyphens from the ends:

<?php
function slugify(string $title): string
{
    $slug = mb_strtolower($title);
    $slug = preg_replace('/[^\p{L}\p{N}]+/u', '-', $slug);
    return trim($slug, '-');
}

echo slugify('PHP preg_replace(): A Quick Guide!'), "\n";
echo slugify('  10 Tips & Tricks for 2026  '), "\n";
echo slugify('Café au lait'), "\n";

Several patterns, a limit and a count

$pattern and $replacement can be arrays: each pattern is replaced by the replacement at the same index, in order. The fourth argument limits replacements per pattern (-1 means no limit), and the fifth receives how many replacements were made.

<?php
$text = 'Call 555-1234 or email me at ana@example.com';

$patterns = ['/\d{3}-\d{4}/', '/\S+@\S+/'];
$replacements = ['[phone]', '[email]'];
echo preg_replace($patterns, $replacements, $text), "\n";

// Only the first match
echo preg_replace('/\d+/', '#', 'a1 b2 c3', 1), "\n";

// Count the replacements
$clean = preg_replace('/\s{2,}/', ' ', 'a  b   c d', -1, $count);
echo "$clean ($count replacements)\n";

Compute the replacement with preg_replace_callback

When the replacement depends on the match, use preg_replace_callback(). It calls your function once per match with the $matches array and inserts whatever the function returns. An arrow function keeps short callbacks on one line.

<?php
// Add 10% tax to every price
$menu = 'Coffee $3.50, Cake $4.00, Tea $2.80';
echo preg_replace_callback(
    '/\$(\d+\.\d{2})/',
    fn($m) => '$' . number_format($m[1] * 1.10, 2),
    $menu
), "\n";

// snake_case to camelCase
echo preg_replace_callback('/_([a-z])/', fn($m) => strtoupper($m[1]), 'user_first_name'), "\n";

// Fill {placeholders} from an array, leave unknown ones alone
$vars = ['name' => 'Ada', 'count' => 3];
$template = 'Hi {name}, you have {count} messages. {unknown}';
echo preg_replace_callback(
    '/\{(?<key>\w+)\}/',
    fn($m) => $vars[$m['key']] ?? $m[0],
    $template
), "\n";

preg_replace_callback_array() takes an array of pattern => callback pairs when several patterns each need their own function.

Replacement text from a variable

A $ or a backslash in the replacement has a meaning, so a replacement that comes from a user or a database can be mangled. A price of $10 is read as a reference to group 10, which does not exist, and comes out empty:

<?php
$price = '$10';

// Wrong: "$10" is read as backreference 10
echo preg_replace('/PRICE/', $price, 'Only PRICE today'), "\n";

// Right: the callback's return value is used as is
echo preg_replace_callback('/PRICE/', fn() => $price, 'Only PRICE today'), "\n";

// Or, for fixed text, no regex at all
echo str_replace('PRICE', $price, 'Only PRICE today'), "\n";

When neither the search text nor the replacement is a pattern, str_replace() is the simpler and faster choice. To test a pattern before replacing anything, see preg_match().