Kitaev's toric code is expected to be at the core of the first generation of quantum computers that will incorporate error protection. To make full use of the toric code, one requires an efficient decoding scheme that will process the classical information obtained from quantum syndrome measurements, so as to be able to regularly put arrays of qubits back into their intended states. The renormalisation decoders introduced by Duclos-Cianci and Poulin exhibit one of the best trade-offs between efficiency and speed. One feature that remained a mystery however, is their behaviour over adversarial channels, i.e. their worst-case behaviour. In this talk, we introduce a relatively natural and deterministic version of a renormalisation decoder and bound its error-correcting radius.