Phone number regex with a live tester

Key facts

Which phone number regex should I use?

Use E.164 to store and send numbers, the lenient US/Canada pattern to check what US visitors type, and the digits-only pattern as a loose check for any country. Each pattern below has a stated basis; per-country patterns are left to libphonenumber, which keeps per-country numbering metadata.

The 7-digit floor on the digits-only pattern is a practical heuristic, not a standard; the 15-digit ceiling is E.164's.

What is the best regex for a US phone number?

For numbers as people type them, use ^(?:\+?1[-.\s]?)?(?:\([2-9]\d{2}\)|[2-9]\d{2})[-.\s]?[2-9]\d{2}[-.\s]?\d{4}$. It accepts (415) 555-0132, 415.555.0132 and +1 415 555 0132, and rejects area codes or exchanges that start with 0 or 1, which the North American Numbering Plan does not allow. Once separators are stripped, the strict form ^(?:\+1)?[2-9]\d{2}[2-9]\d{6}$ is enough.

How do I validate international numbers?

Normalise first, then match E.164. Strip spaces, brackets, dots and dashes, keep the leading plus, and test the result against ^\+[1-9]\d{1,14}$: a plus sign, a country code that never starts with 0, and at most 15 digits in total under ITU-T Recommendation E.164. For per-country rules such as valid lengths and prefixes, use libphonenumber rather than one regex per country.

Which inputs does each pattern accept?

This matrix runs every test number against all four patterns, as typed. "After normalise" means the number fails as typed but passes once spaces, brackets, dots and dashes are stripped. It is generated from the same code as the tester above.

Test numbers use ranges set aside for fiction where one exists: NANP 555-0100 to 555-0199 and the UK's 020 7946 0xxx drama range.

How do I use a phone regex in HTML, JavaScript, Python and PHP?

Normalise the value, then test it against the whole string: an anchored regex in JavaScript and PHP, fullmatch() in Python, and the pattern attribute in HTML, which the browser anchors for you. The snippets below use E.164 for stored values and the US/Canada or digits-only patterns for the input.

<!-- US / Canada: accepts (415) 555-0132, 415.555.0132, +1 415 555 0132 -->
<label for="phone">Phone</label>
<input type="tel" id="phone" name="phone"
       inputmode="tel" autocomplete="tel"
       pattern="(?:\+?1[\-.\s]?)?(?:\([2-9]\d{2}\)|[2-9]\d{2})[\-.\s]?[2-9]\d{2}[\-.\s]?\d{4}"
       title="10-digit US or Canada number, e.g. (415) 555-0132">

<!-- Any country: 7-15 digits; spaces, dashes, dots and brackets allowed -->
<label for="phone-intl">Phone</label>
<input type="tel" id="phone-intl" name="phone"
       inputmode="tel" autocomplete="tel"
       pattern="\+?(?:[\s\(\)\.\-]*\d){7,15}[\s\(\)\.\-]*"
       title="Phone number with country code, e.g. +44 20 7946 0958">

// E.164: "+", a country code, at most 15 digits (ITU-T E.164)
const E164 = /^\+[1-9]\d{1,14}$/;

function normalisePhone(value) {
  // Drop spaces, brackets, dots and dashes; keep the leading "+".
  return value.trim().replace(/[\s().\-\u2010-\u2015]/g, "");
}

E164.test(normalisePhone("+1 (415) 555-0132")); // true
E164.test(normalisePhone("(415) 555-0132"));    // false: no country code

import re

E164 = re.compile(r"\+[1-9]\d{1,14}", re.ASCII)
SEPARATORS = re.compile(r"[\s().\-\u2010-\u2015]")

def normalise_phone(value: str) -> str:
    return SEPARATORS.sub("", value.strip())

bool(E164.fullmatch(normalise_phone("+1 (415) 555-0132")))  # True
bool(E164.fullmatch(normalise_phone("(415) 555-0132")))     # False

<?php
function normalise_phone(string $value): string {
    return preg_replace('/[\s().\-\x{00A0}\x{2010}-\x{2015}]/u', '', trim($value));
}

$phone  = normalise_phone($_POST['phone'] ?? '');
$isE164 = preg_match('/^\+[1-9]\d{1,14}$/D', $phone) === 1;

Why doesn't type=tel validate?

Phone formats vary too much between countries, so browsers apply no format check to <input type="tel">; MDN notes that, unlike type=email and type=url, its value is not validated automatically. type=tel gives mobile visitors a phone keypad, and the only checks come from the pattern, required, minlength and maxlength attributes.

Can a regex tell if a number is real?

No. A regex checks characters and length, nothing more. +14155550132 passes every pattern on this page, yet it sits in the 555-0100 to 555-0199 block NANPA reserves for fictional use. To know a number is assigned and reachable, send a one-time code by SMS or voice call, or use the number lookup API your SMS provider offers.

When should I use libphonenumber instead of a regex?

Use libphonenumber when you accept numbers from more than one country, need to turn whatever people type into E.164, or want a number's type where the numbering plan reveals it. It is Google's open-source library for parsing, formatting and validating international phone numbers, and it carries per-region metadata that a single regex can't. Ports exist for JavaScript (libphonenumber-js), Python (phonenumbers) and PHP (libphonenumber-for-php).

import { parsePhoneNumberFromString } from "libphonenumber-js";

// Parse what the visitor typed, with their country as the default region.
const phone = parsePhoneNumberFromString("020 7946 0958", "GB");

phone?.isValid(); // checks the number against libphonenumber's UK metadata
phone?.number;    // "+442079460958", the E.164 form to store

More phone regex questions

Related

• Email regex tester: the same treatment for email addresses.
• HTML input reference: every input type, including tel, in a working form.
• Server-side form validation: re-check what the browser already checked.
• Contact form with phone number: dial-code dropdown and formatting, ready to paste.
• All free form tools