The aim of thermoeconomic diagnosis of operating energy intensive systems is the determination of fuel consumption, the identification of the causes of its increment from design conditions and the quantification of the effect of each one of these causes. For this task, besides data acquisition and monitoring systems, a thermoeconomic diagnosis methodology is needed. After a review of methodologies, three of them are compared: quantitative causality analysis, linear regression and neural networks. These methods are based on thermodynamic indicators (more close to daily operation) instead of thermoeconomic parameters (which allow a homogeneous formulation); however, the first one is based on a thermodynamic description of the system, while the others are empirical. The comparison is based on the diagnosis of a 3 x 350 MW coal-fired-power plant for a time span of more than 6 years. Results show that quantitative causality analysis is able to quantify the effects of all variables while the others are only suitable for the most influential ones. (C) 2010 Elsevier Ltd. All rights reserved.