Railroads are playing pivotal role to the economic growth of United States and trackbeds ensure their safe and smooth operations. However, reliable trackbed performance prediction has always been challenging due to many reasons, for instance materials characterization, deteriorations of materials and geometries due to railways operation and environmental changes etc. All these factors exhibit varying levels of intrinsic variabilities and uncertainties. These variations and uncertainties are completely ignored in most of the state-of-the-practice problems due to lack of availability of robust models that can characterize variations in materials, geometries, and/or loadings. In this study, a Random Finite Element based three-dimensional numerical model, named ADYTrack, is developed for structural analysis of railroad trackbeds. Uniqueness of this model is the inclusion of materials’ intrinsic variabilities, geometric imperfections and/or uncertainties in axle loadings. The ADYTrack results, when compared with the analytical solution of a cantilever beam model, produced a maximum percentage difference of 0.7%; and 6% difference when compared with ANSYS software results for a single layer trackbed model; and a range of 5-20% difference was observed when validated against the actual field measurements. Sensitivity studies using RFEM based ADYTrack revealed that with the increasing variations in input parameters, measured by coefficient of variations (COV), the variations in output parameter also increased, and generally followed a bilinear trend with first linear component relatively insensitive up to around 30% COV of input parameters. However, beyond this limit, a considerable increase was observed in COVs of output parameters. For a COV of 80% in subgrade resilient modulus, a COV of 65% in vertical stress at the top of subgrade layer was observed. Additionally, the performance of any substructure layer found to be more sensitive to the variations in its own resilient modulus values. Furthermore, resilient modulus of subgrade layer was found to be the most influential input parameter, as revealed by many other studies, and so was its variations. To conclude, ADYTrack model can serve as a robust supplemental tool for railroad trackbed analysis, especially at locations that exhibit higher degrees of uncertainties and thus pose higher risk of public or infrastructure safety.