This work investigates the sensitivity of a superplastic material’s tensile test to the major geometrical parameters of the selected test specimen. This required generating a large number of specimens by systematically varying the gauge length, gauge width, grip length and width of a standard geometry. The specimens were prepared from a moderately superplastic AZ31B-H24 magnesium alloy sheet and were then stretched at a selected rate and temperature. Deformation in each specimen was tracked via an electrochemically-etched fine grid which was particularly used to quantify the amount of material flow from the grip into the gauge region. The consequences of the latter on the accuracy of measured stresses and strains were correlated back to the corresponding geometrical parameters. Ultimately, the results were utilized to set the guidelines for selecting the optimum parameters in a “proper” specimen, for testing the unique class of superplastic materials.