Abstract Systematic laser surface melting experiments have been carried out to investigate the effect of laser-processing parameters on the localized corrosion resistance of sensitized type 321 austenitic stainless steel. The microstructures, as functions of laser-processing parameters, were characterized by means of SEM, TEM, XRD analyses, and the corresponding localized corrosion resistances were evaluated by pitting and electrochemical potentiokinetic reactivation (EPR) tests. Our results showed that the microstructure and the solidification mode in the surface melted layer are very sensitive to the cooling rate. With the increase of cooling rate, the solidification mode will change from primary δ-ferrite to primary austenite. Laser surface melting can effectively eliminate the carbides formed during sensitizing treatment and homogenize the sensitized microstructures, leading to a remarkable improvement in the localized corrosion resistance of the processed material. In order to obtain the optimum technology, a novel artificial neural network method was used to model the non-linear relation between laser-processing parameters and the corrosion resistance, and a genetic algorithm was further introduced to optimize the LSM technology for different property demand. Our verifying experiment indicated that experimental results agree well with the optimized ones. Artificial neural network combined with genetic algorithm offers a new effective means for optimization of laser-processing technology.