IBAN formats and lengths for every country

What is the structure of an IBAN?#

An IBAN (International Bank Account Number) is a string of at most 34 characters in ISO 13616 [4]. The first two letters are the ISO 3166 country code, the next two digits are check digits, and the rest is the BBAN (Basic Bank Account Number), whose layout each country defines.

The registry writes the BBAN layout in SWIFT notation: 4!a6!n8!n means 4 letters, then 6 digits, then 8 digits, all fixed length. n is digits, a is upper-case letters, c is letters or digits. The IBAN length is 4 plus the sum of the BBAN counts, so the United Kingdom’s 4!a6!n8!n gives 4 + 18 = 22.

How does the mod-97 check work?#

ISO/IEC 7064 defines the MOD 97-10 system, which IBAN uses for its check digits [5]. To validate:

1. Move the first four characters (country code and check digits) to the end.
2. Replace each letter with two digits: A = 10, B = 11, … Z = 35.
3. Read the result as one integer and divide by 97. The remainder must be 1.

A remainder of 1 means the check passes. To generate check digits, put 00 in their place, compute the remainder r, and write 98 - r. The number is too large for a 64-bit integer for most countries, so implementations either use big integers or reduce piecewise.

What does a valid checksum prove, and what doesn’t it?#

It proves the string was probably not mistyped. We measured this on the registry examples: of 15,777 single-digit substitutions, none still passed; of 1,140 adjacent digit transpositions, none still passed (method in the Method section). The standard states that these systems detect all single substitution errors and transposition and shift errors [5].

It does not prove more than that:

• Existence. Anyone can compute valid check digits for an invented BBAN. DE63111111111111111111 passes mod-97 and has the right German length of 22; we constructed it, and no account is implied.
• Ownership or status. The checksum says nothing about whether the account is open or who holds it.
• Bank identity. Some countries validate the bank code inside the BBAN with a national scheme; mod-97 on the whole IBAN does not.
• Random matches. A random string with a correct prefix and length passes roughly 1 time in 97. Length plus checksum together are why a scanner can find IBANs in free text with few false positives.

StormDoor’s pii.iban finding uses exactly these two tests, the per-country length and mod-97, on candidates in free text, and also accepts the grouped form in blocks of four (finding page, PII gate). You can test one value in the IBAN checker. StormDoor makes that tool.

Are there countries with partial or non-registry IBANs?#

We could not verify this from the registry, because we could not read it. Two observations from the data we do have: python-stdnum notes in a commit message that it kept Pakistan (PK) in iban.dat because the registry’s PDF still lists it although the text file does not [1], which shows the registry’s formats can differ between its PDF and text editions. And php-iban’s community table lists 121 countries and territories, which includes 36 codes absent from the release 101 data (for example Algeria, Morocco, Iran, and French overseas territories such as Réunion); php-iban marks many of them as not official registry entries [2]. We exclude those from the table. If a country is missing here, check it against the registry before relying on this page.

IBAN formats and lengths by country#

Source: release 101 data via python-stdnum [1]; examples from php-iban [2], each re-checked by our script for country prefix, length, BBAN structure and mod-97. Falkland Islands, Libya, Mongolia and Somalia have no example because our example source did not carry them. Data as of 10 October 2026.