Computational models are widely used in the prediction of occupant injury responses and vehicle structural performance of ground vehicles subjected to underbody blasts. Although these physics based computational models incorporate all the material and environment data, the classic models are typically deterministic and do not capture the potential variations in the design, testing and operating parameters. This paper investigates the effect of one such variation in physical tests, namely, variations in the position of occupant setup on the occupant injury responses. To study the effects of occupant position, a series of vertical drop tower tests were performed in a controlled setup. A vertical drop tower test involves an Anthropomorphic Test Device (ATD) dummy positioned on a seat and the setup is dropped on an energy attenuating surface, thus producing a desired shock pulse on the seat structure. The experimental data was analyzed for sensitivity of occupant position and ATD joint friction variations. Results from this work reiterate the need to include the stochastic variability of test setup and design parameters in the modeling and simulation and pre-test prediction of any physical test including live fire tests. This project work was performed under TARDEC’s NTUBB (Near Term Under-Body Blast) Modeling and Simulation enhancement program.