Weather normalization is often done in building energy benchmarking, so that energy performance of buildings exposed to different weather conditions can be compared. But in this study, it was decided not to do weather normalization because of the following reasons. Firstly, climatic difference across the country is non-existing in Singapore, since it is a small island country with uniform climatic conditions. Secondly, for the benchmarking purpose, energy consumption data of 2004 was used, except for hotels with incomplete data for the year. In the later case, we used energy consumption data of 2005 instead. Unpaired t-test was conducted to compare daily mean outdoor temperature of the two years. The difference was found to be insignificant at 95% confidence level. Thirdly, prevailing weather normalization approaches, such as the degree-day methods, are found not to be very reliable, especially when applied in cooling dominated climates (Akander et al., 2005). Due to these reasons, no weather correction was made to the building EUI. In the future, however, the benchmark will probably be extended to include buildings from neighboring countries like Malaysia, or, building energy consumption data collected from a few years is to be complied in the benchmarking system. When that happens, the necessity of doing weather normalization should be reconsidered, and suitable method may need to be devised.