Recent methods of data envelopment analysis (DEA) deal mostly with static input-output systems. For dynamic systems with both current and capital inputs, the measurement of efficiency poses new problems due to intemporal effects. Some control-theoretic methods are discussed here by way of generalizing the static framework of data envelopment analysis in terms of dynamic multipliers and stochastic parameters. The dual problems of control and state estimation are analysed in their dynamic aspects, which show the myopic nature of measuring efficiency in a static version of the DEA model.