<p>Cost benchmarking, as it is used in public utility regulation, refers to the use of various statistical and non-statistical techniques to estimate the efficient cost function of a regulated firm based on the out-turn performance data of other comparator firms. Whereas benchmarking is often seen as a key part of the toolkit of public utility regulators, there are several fundamental theoretical and practical problems with benchmarking which limit its usefulness as a guide for setting a revenue allowance for a regulated firm. This paper highlights those problems so that practitioners can adopt realistic expectations of what benchmarking can achieve. Whereas benchmarking should continue to be developed to improve its usefulness, these observations set out here suggest caution before relying on cost benchmarking as a primary driver of the revenue allowance of a regulated firm. We propose a ‘code of practice’ that might be adopted by regulators seeking to use benchmarking techniques.</p>