The article shows the effectiveness of neural network modeling for multi-criteria optimization of chemical composition and prediction of strength properties of system-alloyed maraging steels on Fe-Cr-Ni-Mo-based. The analysis of the dependence of the mechanical characteristics of this class of steels on the chemical composition allowed us to determine the effect of small additives of alloying elements on yield strength, strength, toughness and static crack resistance. To solve the problem of multi-criteria optimization of the chemical composition of maraging steels, the “ideal point” method and the linear sequence algorithm were used.