The development and evaluation of cost-effective environmental strategies for the energy sector, based on energy-emission models for both national and international levels, requires consideration of the effects of energy-price changes on energy-service demand. To cope with the problem that the forms of functional dependencies between energy-service demand and independent variables are often unknown, a model based on neural networks has been developed. A multi-layer perceptron (MLP) is proven to be suitable for calculations of price elasticities. The neural network model has been applied to the passenger-traffic sector in the FRG. Elasticities of energy-service demand with respect to price level and direction of price change are derived.