Regex for phone numbers: US formats, E.164, and when to stop

Ask ten people for their phone number and you'll get it back in ten shapes: 555-123-4567, (555) 123-4567, 555.123.4567, +1 555 123 4567, 5551234567. They're all the same number.

A phone regex has to accept the shapes people actually type and still catch the obvious mistakes. This post builds a US pattern one piece at a time, then covers the international format, and ends with the part a regex can't do. Every example is live, so edit the patterns and watch the lines change.

01Start with the digits

A US number is a 3-digit area code, a 3-digit exchange and a 4-digit line number. The strictest version accepts exactly one format:

That's fine if you control the input mask. If people type freely, it's too strict.

02Allow the common separators

Replace each dash with [-. ]?: a dash, a dot or a space, and the ? makes it optional so plain digits work too.

One side effect: each separator is chosen on its own, so 555-123.4567 passes too. Usually that's harmless. If you want both separators to be the same, capture the first one and repeat it with a backreference (more on those in our backreferences guide):

03Parentheses around the area code

The tempting fix for (555) 123-4567 is to make each parenthesis optional with \(? and \)?. That accepts unbalanced input:

Alternation fixes it. Either the area code is in parentheses (with an optional space after), or it's bare (with an optional separator after):

04The optional +1

The US country code is 1, written as +1, 1 or left off entirely. Wrap it in an optional group at the start:

That's a solid pattern for a US-only form.

05Tighten it with area code rules

Real US area codes and exchanges (the middle three digits) never start with 0 or 1. Swapping \d{3} for [2-9]\d{2} in those two spots rejects the classic fake number:

Whether that's worth it depends on your users. It catches a few typos, and it also makes the pattern harder to read.

06International numbers: E.164

Outside the US, formats vary too much for one pattern. Instead of trying, most systems store numbers in E.164, the international standard: a +, the country code, then the number, with no spaces or punctuation and at most 15 digits. Country codes never start with 0.

The usual approach is to let people type their number however they like, strip the punctuation, and then check the result. A regex is perfect for the stripping step:

In Python that's re.sub(r"[\s().-]", "", number). In JavaScript, number.replace(/[\s().-]/g, "").

07What a regex can't tell you

^\+[1-9]\d{1,14}$ checks the shape. It can't tell you:

• whether the country code exists (+999 isn't one, and the pattern accepts it),
• whether the length is right for that country (UK numbers and German numbers have different lengths, and some countries mix lengths),
• whether the number is a mobile, a landline or unassigned,
• whether anyone answers.

For real validation, use libphonenumber, Google's phone number library. It knows the numbering plan of every country, and it has ports for most languages: phonenumbers in Python and libphonenumber-js in JavaScript. It also formats numbers for display and converts them to E.164 for storage:

import phonenumbers

n = phonenumbers.parse("(212) 555-0199", "US")
phonenumbers.format_number(n, phonenumbers.PhoneNumberFormat.E164)  # '+12125550199'
phonenumbers.is_possible_number(n)  # True: the length fits the US plan

And like with email addresses, the only way to prove a number works is to send a code to it.

08Try one

Here's a board of phone numbers in mixed formats. You don't need to check every format separately. Every monster has ten digits in a row (ignoring the punctuation), and no innocent does:

09Checklist

• For US-only input, ^(\+?1[-. ]?)?(\(\d{3}\) ?|\d{3}[-. ]?)\d{3}[-. ]?\d{4}$ covers the common formats.
• Use alternation, not two optional parentheses, so ( and ) come as a pair.
• For international numbers, strip punctuation with [\s().-] and store E.164.
• Check the shape with ^\+[1-9]\d{1,14}$, and use libphonenumber when it matters.
• Send a code if you need to know the number is real.

10Keep practising

• The Real-world validators track in the Academy covers ZIP codes, dates, times, emails and more, one hunt at a time.
• The common patterns section of the guide has more live examples.
• Try today's hunt, or work through older boards in the archive.