Mass yield and operating time for a pork meat dehydration-impregnation-soaking (DIS) process were optimized using a coupled genetic algorithm/sequential quadratic programming method in order to obtain the optimal operating conditions: temperature and soaking solution concentrations. The DIS process was simulated by a neural network model. The non-linear optimization problem was constrained to ensure the main product characteristics: stability indicated by the water activity target and flavour characterized by the phenol gain target. The climatic conditions, the model validity region, the raw material costs and the operator working schedule were taken into account. Optimal solutions are discussed for three different batch configurations: single-stage processing under constant conditions, singlestage processing under varying temperature and two-stage processing under constant conditions. The most convenient operation resulted in a two-stage soaking process because of time, energy and cost savings, control convenience, product cooling anticipation and a reasonably high mass yield.