The building energy simulation computer program DOE-2 was used to carry out a parametric study of a generic high-rise air-conditioned office building in Hong Kong. A total of 28 design parameters related to the building load, the heating, ventilating and air-conditioning (HVAC) system and the HVAC refrigeration plant were found to correlate well with the predicted annual electricity consumption. Both linear and non-linear multiple regression techniques were used to develop regression models and energy equations for the prediction of annual electricity use. Twelve input design parameters (six from building load, four from HVAC system and two from HVAC refrigeration plant) were considered to be the most significant design variables and were used in the energy prediction equations. It is hoped that the resulting models and energy equations can be used as a simplified design tool for comparing the relative thermal and energy performance of different design schemes during the early design stage. This paper presents the methodology and the development work, and discusses the findings.