The purpose of this study is to predict and optimize the internal combustion engine performance using diesel-natural gas fuel using the artificial intelligence. The ultimate goal is to determine the optimal substitution rate of natural gas to minimize the costs of operation and pollutants emissions such as carbon monoxide CO, hydrocarbons HC and nitrogen oxides NOx, considering the values of efficiency. The analyzed data are obtained from tests performed earlier. The procedure involves training, validation and test (using neural networks). Once these data were analyzed with different trained neural networks for learning and prediction, which are maps of the predicted values based on experimental data have been seized. Finally, and continuing with the process of optimization (technique of Genetic Algorithms), is given the best substitution rate of and lower emissions in the maps generated. The results indicate a good agreement between data and neural network, the process of optimization using certain items of work appropriate for each case analyzed.