This paper presents an optimization model of the power generation potential for either new or repowered hydroelectric plants. It is based on curves that represent the unit efficiency as a function of the nominal output. The objective is to choose the combination of efficiency curve types that maximizes the power generation for certain load levels. The mathematical formulation results in a mixed integer, nonlinear programming problem. Genetic Algorithm is employed to solve this. The operators and parameters of the model are chosen by simulation using the objective function values as a selection method. A case study is carried out for two Brazilian hydroelectric plants: Sobradinho and Ilha Solteira. The results show the importance of the turbines model choice in order to get the maximum benefit of a plant.