A new Bayesian approach is proposed to estimate the annual energy production (AEP) of a site where construction of wind turbines is considered. The approach uses long-term wind speeds of a nearby weather station and short-term wind speeds near the target site. Uncertainties exist due to the limited amount of data in the target site, in addition to the inherent uncertainties in the wind speed, the air density, the surface roughness exponent, and the power performance of the turbine. The proposed method systematically addresses these uncertainties and provides the distribution of the AEP. For illustration, we used the wind speed data near Yeosu, Korea, and the power performance curve of a 3 MW turbine. For the site and the turbine studied, the range given by the 95% confidence interval corresponded to 8.9% of the mean AEP, and the range given by the 99% confidence interval corresponded to 11.9% of the mean AEP. Benefits of using the Bayesian approach compared to the classical statistical inference was also illustrated with the case study. The proposed approach provides a more conservative estimation considering the uncertainties due to the limited amount of data. Distributions of parameters of the prediction model are also provided, which enables a more detailed analysis of the prediction. (c) 2012 Elsevier Ltd. All rights reserved.