The decay of existing power plants is caused primarily by mechanical factors, human errors, environmental effects, control deviations, and defects in materials. These causes are reflected in performance malfunctions of the processes. In a highly integrated and complex energy system, it is difficult to detect and evaluate the origin of a malfunction. A diagnosis methodology for power generation systems which addresses this is presented here and illustrated by its application to the TADEUS problem (Valero et al., 2004, 2002a). The proposed methodology is based on a comparison between two operating conditions, namely, the test operating conditions (TOP), which reflect actual operating conditions, and the reference operating conditions (ROP), which can be based on simulated thermal balances or an acceptance test. Energy and exergy balances as well as analytical models are used to simulate plant performance at either TOP or ROP and these, along with data collection from the plant, characterize a complete database for the plant. Such a database includes variables which can be classified as either free (variations in their values are the cause of malfunctions) or dependent. A reconciliation-simulation algorithm corrects the values of the free variables for a given heat rate and total power to pre-malfunction values. In order to demonstrate the reliability of the approach, it is applied to the combined cycle power plant outlined in the TADEUS problem. It is shown that the results generated by this methodology provide relevant information of the present condition of the plant.