The manufacture of open cell metal foams by dissolution and sintering process (DSP) is the matter of the present work. Aluminum foams were produced by mixing together carbamide particles with different mesh sizes (i.e., space-holder) and very fine aluminum powders. Attention was first paid at understanding the leading phenomena of the different stages the manufacturing process gets through: Compaction of the main constituents, space-holder dissolution, and aluminum powders sintering. Then, experimental tests were performed to analyze the influence of several process parameters, namely, carbamide grain size, carbamide , compaction pressure, and compaction speed on the overall mechanical performance of the aluminum foams. Meaningfulness of each operational parameter was assessed by analysis of variance. Metal foams were found to be particularly sensitive to changes in compaction pressure, exhibiting their best performances for values not higher than 400 MPa. Neural network solutions were used to model the DSP. Radial basis function (RBF) neural network trained with back propagation algorithm was found to be the fittest model. Genetic algorithm (GA) was developed to improve the capability of the RBF network in modeling the available experimental data, leading to very low overall errors. Accordingly, RBF network with GA forms the basis for the development of an accurate and versatile prediction model of the DSP, hence becoming a useful support tool for the purposes of process automation and control.