We define environmental performance indicators as analytical tools that allow one to compare various plants in a firm, or various firms in an industry, with each other and with respect to certain environmental characteristics. We start from a reflection about the kind of information that should be incorporated into such indicators, and the extent to which that information should be aggregated. Existing approaches are both rare and dissimilar. They range from oversimplified indicators to more sophisticated ones, in which there is a trend to take somewhat arbitrary viewpoints. In the scope of the theory of productive efficiency, three categories of factor are taken into account, i.e. inputs, desirable production outputs and pollutants in the form of ''undesirable'' outputs. Non-parametric efficiency measures easily and usefully lend themselves to the derivation of environmental performance indicators. They are the duals of indicators that can be obtained in the traditional framework of data envelopment analysis. We use data envelopment analysis to define standardised, aggregate environmental performance indicators, that is, quantities comprised between 0 (bad performance) and 1 (good performance). Such indicators do not require the specification of any a priori weight on the environmental impacts that are being aggregated. A discussion is proposed on such topics as an analysis of the nature and causes of environmental inefficiencies, and the relationship between ''environmental performance'' as defined in this paper and the actual global effect of industrial activities on health (toxicity) and the environment.