Estimation of accurate in-service life is of great imp0l1ance to the power generating industries, especially thermal and nuclear power plants. Since only a small amount of material is available for testing purposes. a miniature test is found to be of immense utility. In this study. six hundred and sixty finite element simulations of the small punch test were car­ried out considering different material and frictional param­eters. Based on these results, five neural network models were developed. Successfully trained artificial neural net­works were used to predict flow properties and yield stress. The m1ificial neural network results were finally validated with the experimental results and the two most suitable models were selected. The proposed approach offers a powerful reverse engineering tool for material characteriza­tion.