Recently the non-destructive test technique which uses the grain boundary etching characteristics owing to the variation of material structures has been proposed. However, during in-serviced GEM test there are a lot of variables such as the changes of temperature and concentration of etching solution, the roughness condition of surface polished etc.. The purpose of this paper is to investigate the influences of these test variables on GEM test results in order to establish a reliable and sensitive of GEM evaluation technique. The experiments are conducted in various solution temperatures, 10$^{C}$, 15$^{C}$, 20$^{C}$, and 25$^{C}$ and in 70% and 100% concentrations of that, and in various surface roughnesses polished by #800, #2000, and 0.3㎛ alumina powder. Through the test with variables, it is verified that the decrease of temperature and concentration of etching solution and the coarsened surface roughness by not using polishing cloth and powder induce some badly and/or greatly influences on GEM test results like grain boundary etching width(W$\_$GB) and intersecting point ratio(N$\_$i/N$\_$0/). Therefore, to get reliable and good GEM test results, it must be prepared the surface of specimen polished by polishing cloth and 0.3㎛ alumina powder and the saturated picric acid solution having 25$^{C}$ and be maintained the constant temperature(25$^{C}$) during GEM test.