Measuring energy efficiency is crucial when planning energy reduction policies. However, decision makers are understandably reluctant to act in the absence of solid data and results supporting a policy position. This paper proposes a methodology to measure the energy efficiency of farms based on the Data Envelopment Analysis (DEA) approach. In a manner similar to the cost framework, we decompose energy efficiency measurements into technical and allocative efficiencies in energy contents of inputs. In this study, we replace input prices used in traditional economic efficiency measurements by their energy content. We use the energy efficiency model to explore the optimal input-mix that produces the current outputs at minimum energy-consumption. We demonstrate that this decomposition can help policy makers design accurate energy policies. However, the uncertainty of the data and, more particularly, the energy content of the inputs leads us to recommend using the methodologies to calculate the bounds of efficiency to obtain more plausible and robust results. Based on our analysis, energy efficiency is not a fixed value, and policy-makers should consider it with caution. We use a 2007 database of French farms specialised in field crops for empirical illustration.