Assessing the sustainability of a regional agricultural system is complex, because in different decision-making contexts stakeholders can use different criteria and methodologies, thus arriving at different and contrasting judgments. One way of dealing with the complexity of measuring the concept of sustainability is to adopt a multidimensional perspective, which recognizes the presence of an economic dimension which requires feasibility, a social dimension which requires acceptability, and an environmental dimension which requires carrying capacity. Many approaches for measuring sustainability face the difficulty to reconcile this multidimensional perspective with the necessity to come up with a "synthetic" and one-dimensional assessment measure that could be used for both policy-making and methodological purposes. The goal of this paper is to contribute to the development of a methodological approach that can simplify the assessment procedure of sustainability of agricultural systems, while considering the multidimensional perspective. We used the three dimensions of sustainability to define two economic indicators, two social indicators, and four environmental indicators of sustainability. Then we used Data Envelopment Analysis (DEA) to partition 252 European agricultural regions into a subset of DEA-efficient regions and a subset of non-efficient regions under five scenarios. The scenarios reflect preferences with respect to the importance of the three dimensions of sustainability. Impact of model choices such as constant versus variable returns to scale, input versus output orientation, and balancing constraints is shown. The combination of multidimensional perspective and DEA allowed to operationalize the complex and sophisticated concept of sustainability. Applying DEA at the EU regional level enabled analysis of the heterogeneity of performances within each EU Member State and among them. This heterogeneity is a fundamental research topic in the domain of assessment of sustainability of agricultural systems. (C) 2013 Elsevier Ltd. All rights reserved.