A theoretical-experimental methodology was developped in this work for evaluating the performance of cogeneration plants, including the data comparison with manufacturer specifications, uncertainty analysis of main parameters and thermoeconomic analysis. As a result, this work aims the establishment of an optimized criteriumfor especifying and operating cogeneration systems. Two existing cogeneration plants were analyzed in this work, (a) Companhia Cervejaria Brahma, located in Campo Grande, RJ, consisting of three 4,90 MW gas turbine generators, three heat recovery boilers, including after burners, with a nominal capacity of 36000 kg/h of steam, each one, and (b) Parque Gráfico do InfoGlobo, located in the Wshington Luis Highway, Duque de Caxias, RJ, consisting of two 2,90 MW reciprocating engine gas generators, two heat recovery boilers and 800 TR absorption chiller. A chemical reaction mass balance methodologywas used in the BRAHMA plant.It uses the measurement of pollutant emissions and natural gas compositionfor estimating the turbine inlet air and exhaust gas flow rates, which are important for evaluating the plant performance. Thei methodology was preferred due to the fact that no air and exhaust gas flow rate measurement instrument was installed in the plant, which is usually the case. This methodology was also used for evaluating the performance of heat recovery boilers with after burners. An uncertainty analysis procedure was developed and used to identify operational problems like fouling, reducing the effectiveness of the heat recovery economizer. One of the main results of this work was the fact that as least a 6 percentual point difference between the gas and the steam sides was measured for the overall fuel chemical energy usage, demonstrating the need of a more careful analysis of component performance for designing and specifying a cogeneration plant, which takes into account the heat losses. Thus, one suggests that the heat transfer to the water be specified, rather than the one from the hot gases. A technical-economic analysis of the plant was carried on, and its cost-benefit determined. Due to difficuties in obtaining experimental data, a simulation procedure had to be used for analyzing the performance of the InfoGlobo plant. Design data for several equipments were used in the calculations, supposing that they do not vary too much with ambient conditions. Thus, the uncertainty analysis was not carried on. The used methodology identified the fact that the boiler feedwater pump was not operating in the design point. In a general way, it was observed that the overall fuel chemical energy usage varies too much along the day, indicating a mismatching between load and energy supply. Finally, as a tool for the plant analysis, a computational model was developped for estimating the cogeneration plant component parameters, to be used for design purposes. Partial load operation of the turbines is contemplated in this model, as a function of ambient conditions.