The objective of this paper is to identify the design variables of the compartment that are most influential in determining the extent of damage due to in-compartment airborne explosions (INCEX). From a comparison of numerical simulation results using the CONventional Weapons Effects Program (CONWEP) with results from airborne blast experiments, CONWEP generated blast pressures with reasonable accuracy. From a comparison of numerical simulation results using the Horsford-Coulomb and local necking hybrid fracture model (HC-LN model) with results from indentation experiments, HC-LN model accurately predicts steel fractures. The engine compartment was selected for the damage variable sensitivity analyses in terms of the probabilities of being hit, flooded, and incapacitated. When three TNT masses, two stand-off distances, and eight engine room dimensions with six levels of each dimension were considered, more than 10 million INCEX cases were required. The Latin hypercube sampling technique was adopted to reduce the number of INCEX simulations to 5000. The element-based and image-based methods were applied to evaluate the damage extents. The element-based method underestimated and overestimated the damage area and damage perimeter more than the image-based method, respectively. The most influential design variables on the damage extent were bulkhead thickness and curtain plate height, respectively.