Energy intensity of manufacturing varies due to the products produced and the efficiency of production. For some sectors a portion of the energy demand is weather sensitive. This paper applies stochastic frontier regression to auto assembly and pharmaceuticals. The effects of production, utilisation and weather on energy are estimated. The differences in efficiency are captured by a one-sided error term in the frontier model. Confidential plant level data provided by companies that voluntarily participated in this study are used. US EPA Energy Star programme uses the models to provide recognitionof superior energy performance for plants in these industries. © 2009 Inderscience Enterprises Ltd.