In this study we introduce a new idea of utilizing algorithms from the Computational Intelligence community in building accurate models for saline water evaporation rates. Three experimental methods were used to measure the evaporation rate for different brine concentrations, different water and air temperatures, and different air velocities. A large set of experimental data was collected and then used in creating these models. Two algorithms were applied in the learning process: neural network (NN) with a gradient-descent algorithm, and a hybrid system composed of NN trained by a genetic algorithm (GA). Each algorithm was allowed to use the same training time. The resulting models show excellent accuracy compared to the state-of-the-art models existing in the literature.