Ethereum’s ERC-55 checksum does not require every encoded address to contain both uppercase and lowercase letters. The standard’s own test cases include all-uppercase and all-lowercase results alongside mixed-case examples. What matters is whether each letter has the case selected by the checksum calculation. Judging an address by its visual mix of capitals would miss that distinction.
| What you need to know |
|---|
| ERC-55 encodes a checksum through the case of hexadecimal address letters. |
| The calculation hashes the lowercase hexadecimal address string with Keccak-256. |
| Digits remain unchanged because they have no uppercase form. |
| The official test cases include all-uppercase, all-lowercase and mixed-case results. |
The official ERC-55 specification starts with a 20-byte address and converts it to hexadecimal. It hashes the lowercase hexadecimal text, then uses that hash to select the case of each letter. In the supplied implementation, a corresponding hash digit of eight or higher makes a letter uppercase; otherwise it stays lowercase. Decimal digits are copied unchanged. The result retains the 40-character hexadecimal address length, with the familiar 0x prefix added by the implementation. Capitalization carries checking information without adding a separate suffix.
Read the test cases before rejecting a format
The document explicitly groups its examples under “All caps”, “All Lower” and “Normal”. Those are test vectors for this encoding, not a suggestion that any string written entirely in one case has passed a checksum. The distinction is between an exact result produced by the algorithm and a broad visual category. To test this particular encoding, compare the submitted text with the computed result; spotting a capital letter is not that comparison. Likewise, manually lowercasing a mixed-case result changes the checking information, even though it leaves the hexadecimal address value unchanged.
Check the encoding, then check the destination
ERC-55’s calculation takes an address as its input. It does not ask who controls that address or why a payment is being requested. Passing this format check therefore cannot establish that the destination belongs to the person you intend to pay. For a hardware-wallet receiving address, the separate device-screen comparison answers a different question: whether the destination matches the address displayed by the device. Trezor’s trusted-display guidance explains why an address shown on a computer needs that separate check. Keep both results specific. One checks the address’s case encoding; the other compares a destination with an independent display. Neither requires inventing a universal rule that a correct checksum must look mixed-case.






