Data Envelopment Analysis (DEA) is widely used to compare the empirical performance of decision making units such as firms or public bodies. With a view to the recent boom of applications to a very wide range of different problems, this paper shows for which data and problem structures DEA represents an appropriate analytical tool for generating knowledge. The paper stresses the point that an efficiency analysis such as DEA which does not use prices to weigh different inputs and outputs is, first of all, adequate if several input and/or output prices are unknown or if they cannot be "trusted" to reveal the relative scarcity of resources and goods. In all other circumstances, stand-alone DEA generates less knowledge than standard approaches which resort to classical economic performance criteria that measure the resources used (inputs) and the goods and services produced (outputs) in monetary units. If prices are known, DEA is only useful if it is used as an additional tool to decompose economic performance weaknesses into a technical component (mismanagement of technological processes) and allocative components (maladjustments to prices on the factor and product markets).
