The estimation of energy crop yields is important, to help the firms responsible for collecting them to estimate biomass production in a given area, for example. A Bayesian modelling framework for site- specific yield estimation is presented in this paper. The proposed approach is based on a hierarchical model describing between-site and within-site yield variability. Probability distributions are used to describe the uncertainty of model estimations. The model can be fitted to site-specific yield data, to obtain both average and site-specific yield estimates. Site-specific yield data may be obtained from measurements for crop species other than those for which estimations are required, or from past measurements on perennial crop species grown over a period of several years at a given site. These two options were illustrated in two case studies, in which our model was used to estimate the yields of several energy crops. In most situations, site-specific yield estimations were more accurate than average estimations.