In the United States, railroads have been well established over several decades to meet the needs and demands of transporting freight and passengers. Given the mature rail network, the network's enhancement is often done by reclassing or repurposing existing rails or acquiring new rails. This thesis focuses on optimizing the topology of a network and proposes a methodology to optimize its efficiency using track reclassing, acquisition, and repurposing as means for topological changes. The network efficiency is selected as the primary network attribute. Due to the computational burden associated with computing network efficiency, this study proposes the use of the standard deviation of the node degree as an approximation of network efficiency in identifying optimal solutions. The approximate solution produces results reliably with computational efficiency and accuracy. A case study of a single Class I rail network is introduced to compare the solutions of these two optimization criteria. The results show that the standard deviation of node degree can be used to obtain an optimal solution and offers an adequate and more computationally efficient approximation than the direct use of network efficiency with differences less than 0.2% based on adding 40 links.