Events during last two decades have shown that railways have become the primary interest of terrorist activities.  But, planning for protection against these threats on railways is challenging because of their geographical extensiveness, open and easily accessible infrastructures. Hence, identification and protection of only the critical components of railway infrastructures are very essential. In this paper, we present an optimization model to allocate the protective resources among the components of the railway systems optimally to minimize the impact of worst-case disruptions. The main focus is to enhance the applicability of the model by capturing additional realism and complexities of railway systems. We formulate the problem as a tri-level mixed-integer program. Then, we simplify it a bi-level problem and solve it by super valid inequalities based decomposition algorithm. Computational results of a case study on South East Coast Indian railways are analyzed for efficient protection decisions.