The hydrodynamic performance of a gas-stirred ladle is analyzed experimentally, through a factorial experimental design, through physical modeling and particle image velocimetry techniques, to find the effects of the nozzle position, and its number, the gas flow rate and the presence or not of an oil layer on the flow velocity field, the turbulent kinetic energy distribution, and the open eye area, A. Flow patterns changed drastically in the presence of one or two plugs and as a result of the nozzle’s positions. The turbulent flow fields showed significant differences under the presence or absence of an oil layer and the open eye area strongly depends on the gas flow rate, the position and the number of nozzles. Responses , , and A were statistically processed to identify the relevant effects of each variable on , , and A. These responses were used to calculate the optimum process conditions that minimize A and maximize both and , and then improve mixing efficiency in ladles using multi-objective optimization through the genetic algorithm (GAmultiobj). A Pareto characterization and its analysis were performed to reveal the embedded operation rules allowing optimum performances of the ladle for specific production goals.