ADEME (the French national environmental and energy agency) develops tools in order to measure farm energy performance. The actual measurement is based on the total amount of energy consumed by farmers. The main objective of this paper is to propose an alternative method that can be used in order to improve this measurement. The alternative method that we propose is based on Data Envelopment Analysis (DEA) models. Following the procedure adopted in a cost framework by Farrell (1957) and developed by Färe et al. (1985), we propose to decompose an overall energy performance measurement into two components, namely technical and allocative performances. In order to do this, we replace prices by energy content of inputs. We show that this decomposition can considerably help policy makers to design accurate energy policies. The presence of uncertainty on data, and more particularly on energy content of inputs, leads us to recommend exploiting the methodology proposed by Camanho and Dyson (2005) in order to produce more robust results. Thus, this methodology allows deriving both upper and lower bounds for the performance measurements. A year 2007 database of French farms specialized in crops is used for empirical illustration.