This paper is to review what kinds of mathematical methods used in developing benchmarking systems, to discuss the properties of the methods, and to classify two kinds of benchmarking systems based on their properties. We find that while benchmarking systems are developed by using the energy-use performance of a significant number of reference buildings, benchmarking results can be used to encourage poor reference performers (in energy-efficiency) to improve their performance. On the other hand, because benchmarking systems also function as a public yardstick of energy-use performance in buildings, some regulators release benchmarking information to the media. This proves advantageous because it brings public pressure on owners/developers of poorly performing non-reference buildings. However, not all benchmarking systems can be used by public users (i.e., other non-reference building owners). Depending on whether the resulting benchmarking system can be used in public, we note that there are two kinds of benchmarking system: public benchmarking and internal benchmarking. These two types of benchmarking system are developed by different methods.