Material response under dynamic crash loading is an essential aspect for automobile design to assess vehicle crashworthiness and improve safety performance. However, due to challenges in instrumentation, mechanical property characterization at strain rates of 0.1-10 3 s À1 is not standardized, although these conditions are most relevant to automobile collision scenarios. [1][2][3][4] First, in order to perform high-speed mechanical tests in this strain rate regime, the testing machine actuator is required to have speed capabilities on the order of several hundred in s À1 and necessary dynamic load capacity of several thousand pounds. Second, upon high-speed loading, load cells used for force acquisition exhibit significant ringing and the noise in the signal due to wave propagation, inertia effects, and the frequency response of the transducers. [1,3,[5][6][7][8] This often results in a great deal of uncertainty in stress analysis. Third, accurate strain measurements in the gage section become more difficult as strain rate increases. The most common method to estimate strains is based on the changes in actuator displacement, i.e., through the linear variable differential transformer (LVDT). However, LVDT is not accurate at high speed due to their limited frequency response. Further, LVDT measurements consist of displacements of the loading train and deformations of the entire specimen, hence the contributions of the machine/ loading train compliance to the outputs. [3] Another popular technique for direct measurement of strains of using resistance strain gages is successful other than the limitations of localized measurements and limited measurement range. Recent advances in non-contact optical displacement/strain measurement methods are a solution to the challenges encountered by LVDT and strain gages. [9][10][11][12] Digital image correlation (DIC) as one of the successful technique among the COMMUNICATION Characterization of the material mechanical behavior at sub-Hopkinson regime (0.1 to 1 000 s À1 ) is very challenging due to instrumentation limitations and the complexity of data analysis involved in dynamic loading. In this study, AZ31 magnesium alloy sheet specimens are tested using a custom designed servo-hydraulic machine in tension at nominal strain rates up to 1 000 s À1 . In order to resolve strain measurement artifacts, the specimen displacement is measured using 3D Digital Image correlation instead from actuator motion. The total strain is measured up to % 30%, which is far beyond the measurable range of electric resistance strain gages. Stresses are calculated based on the elastic strains in the tab of a standard dog-bone shaped specimen. Using this technique, the stresses measured for strain rates of 100 s À1 and lower show little or no noise comparing to load cell signals. When the strain rates are higher than 250 s À1 , the noises and oscillations in the stress measurements are significantly decreased from % 250 to 50 MPa. Overall, it is found that there are no significant differences in the elo...