Variable injection-pressure can be used to reduce emissions and increase engine Power. Experiments have been performed on a turbocharger diesel engine for four different injection pressures at three engine throttle positions. The measurements values of the engine Power and NO., emission have been investigated. In this investigation, group method of data handling (GMDH)-type neural network and evolutionary algorithms (EAs) are used for modeling of the effects of the engine speed (N), throttle-position (TP) and injection-pressure (IP) on both engine Power and NOx emission using data provided in the experiments. Employing the obtained polynomial models, multi-objective EAs are then used for Pareto-based optimization of the engine considering two conflicting objectives (engine Power and NOx).