The use of hybrid power generation systems is an attractive alternative to conventional fossil fuel generation since they may assist in mitigating the emission of gases that are harmful to the atmosphere when using clean and renewable sources of energy. However, finding the ideal configuration for the installation of a hybrid system composed of solar photovoltaic (PV)-diesel generation is a complex task. In this sense, the objective of this study is to develop an approach to select the optimal configuration of hybrid power generation systems for isolated regions by means of combining the techniques of Mixing Design of Experiments, Normal Boundary Intersection and analysis of super efficiency using Data Envelopment Analysis. The proposed approach is applied to a set of four isolated regions in the northern region of Brazil, more specifically in the state of Amazonas. The results show that for each region a different configuration is selected but with large shares of diesel generation at first. On the other hand, all these cases represent points in the Pareto frontier that are the most inefficient due to the high volume of CO2 emissions. From the application of the proposed approach, significant CO2 emission reductions are obtained by selecting the optimal configurations represented as the most efficient points in the Pareto frontier. Our results show that due to conflicting characteristics of the selected objectives, the installation of such hybrid power generation systems produces an increase in LCOE, mainly related to the high costs of the batteries, although less accentuated than the reductions in emissions.