Building energy models provide valuable insight into energy use in commercial and residential buildings based on architecture, materials and thermal loads. They are used in the design of new buildings and retrofitting to increase the efficiency of older buildings. The accuracy of these models is crucial to reducing energy use in the US and building a sustainable energy future. In addition to the architecture and thermal loads, building energy models also must account for the effects of the building's occupants on energy use. Traditionally simple schedule based methods have been used to account for the effects of occupants. However, newer research has shown that these methods often result in large differences between the modeled and actual energy use of buildings. In this paper we discuss building energy models and their accuracy in predicting energy use. In particular we focus on the different types of validation methods which have been used to investigate the accuracy of building energy models and how they account for (or do not) the effects of occupants. We also review newer work on stochastic methods for estimating the effects of occupants on energy use and discuss the improvements necessary to increase the accuracy of building energy models. (C) 2012 Elsevier B.V. All rights reserved.