Measuring building energy consumption through surveys is necessary for building energy conservation. It is not feasible to survey all buildings in a given area, but statistical theory can be used to improve the representativeness of survey data. This approach was used to establish survey parameters, develop a sampling method, and calculate sample size. Experimental sampling data is often difficult to obtain, so simulations were used as a substitute for survey sampling. It was determined that the simulation results were consistent with the recorded data. In order to determine the appropriate sample size for large numbers of buildings that cannot be surveyed directly, the relationship between sample size and the total number of buildings was analyzed using four Chinese cities as models. The results showed that as the total number of buildings increased, sample size increased as well, and the threshold at which sample size no longer changed indicated the appropriate sample size. (C) 2011 Elsevier B.V. All rights reserved.