Natural variability, commonly observed in blast and ballistic testing, is not captured in traditional, deterministic Finite Element Analyses (FEA). Sources of variability exist within both the threat and the target and include differences in material properties and geometric features due to the manufacturing process. Variability in the impact conditions (e.g. yaw and angle of impact) can also be a key factor and lead to significantly different outcomes. In this initiative, Dstl has generated a probabilistic FEA-based modelling approach that incorporates the aforementioned variability. This process is based on the methodologies outlined in the probabilistic analyses approach published by the National Agency for Finite Element Methods and Standards. Sensitivity studies have been conducted to identify those ‘key parameters’ which exert the greatest influence on the outcome of general threat/target interaction. The statistical distributions associated with the key parameters have been applied within FEA using the LSTC/ANSYS LS-OPT software package. Statistical distributions of the impact variables are taken from a similar ballistic test series conducted by Lynch et all, which comply with STANAG 4164 and STANAG 2920. Statistical distributions associated with the other parameters have been taken from material characterisation tests and manufacturing standards. This work focuses on the natural variability associated with the impact of Fragment Simulating Projectiles into Low Carbon Steel plate. Initially baseline deterministic analyses, with no variability, were validated against existing ballistic test data. Then, based on the identified statistical distributions, each input parameter was adjusted in isolation and simulated deterministically. The residual velocities from these simulations were compared to the baseline deterministic analyses, allowing the effect of the input parameters to be interrogated. This study has provided a greater understanding of the input parameters that have the highest influence on ballistic events. The presented work provides a foundation for probabilistic modelling in ballistic applications, including numerical V50 assessment for metallic plate-like structures and numerical-generation of ballistic performance response surfaces for implementation within vulnerability and lethality codes.