ENS Namehash Calculator

ENS doesn’t look up vault.finance.eth as a string anywhere on-chain. Every name a resolver contract stores is first collapsed into a single 32-byte value called its node, and that collapse is the namehash algorithm from EIP-137. It isn’t one hash of the whole string. It processes labels right to left, folding each one into the last result, which is why finance.eth and vault.finance.eth share no structure in their output even though one contains the other.

This tool runs that recursion client-side and shows every intermediate node, not just the final one, because the right-to-left order is the detail people get backwards when computing it by hand or copying pseudocode without checking the loop direction.

Worked Example: A Two-Label Subdomain

Click Load Sample and the input fills with vault.finance.eth. The tool splits it into eth, finance and vault, processes them in that order starting from the 32-byte zero node, and the breakdown panel shows each label’s own Keccak-256 hash alongside the running node after it’s folded in. The node after eth is 0x93cdeb708b7545dc668eb9280176169d1c33cfd8ed6f04690a0bcc88a93fc4ae, the same value every .eth name shares as its parent, and the final namehash, after vault is folded in last, is 0x012f568ae19a292705aff770afd1bb534a4565ec446c2602bc314221616137df.

The Algorithm, Written Out

Start with node = 0x00…00 (32 zero bytes). For each label from the rightmost (the TLD) to the leftmost (the subdomain), compute node = keccak256(node + keccak256(label)), where + means byte concatenation, not addition: 32 bytes of the running node followed by 32 bytes of the label’s own hash, hashed together as one 64-byte input. Run out of labels and the last node produced is the namehash. This implementation was checked during development against independently computed reference values for the empty string, a single label, and multi-label names, and matched every one exactly.

Accuracy Limits: Non-ASCII Labels

• Plain ASCII input is lowercased automatically before hashing, since ENS names are case-insensitive but namehash itself hashes raw bytes, so skipping that step would make Vault.eth silently produce a different node than the registered vault.eth.
• An empty input isn’t rejected: it returns the 64-zero-digit node 0x0000000000000000000000000000000000000000000000000000000000000000, ENS’s own root node, exactly as EIP-137 defines it.
• Two dots in a row, like foo..eth, produces an empty label the algorithm can’t hash meaningfully, so the tool reports “Empty label — remove the extra dot between labels.” rather than silently skipping it.
• This tool hashes exactly the bytes you typed. It does not run ENSIP-15 Unicode normalization, so a name built from confusable or mixed-script characters may not match the node ENS’s own resolvers compute for the “same” name after their full normalization pass. Stick to ASCII labels when the result needs to match on-chain resolution.