According to the Canadian Urban Transit Association (CUTA), 140 Billion CAD is required to maintain, rehabilitate, and replace subway infrastructure between 2010 and 2014. The current practice adopted by transit authorities for prioritizing subway stations for rehabilitation is based on the station structural needs. While this classification is reflective of station condition, other factors, such as station size, location and passenger capacity, play an important role. The criticality of a station is an index that represents the functional importance of a station depending upon a set of identified factors. The system criticality is based on several attributes, such as station location, size, and nature of use. This paper presents a novel method of clustering subway stations for rehabilitation priority based on their criticality level. The different stations in a subway network are rated according to their relative importance against predefined attributes. The weights and scores of the attributes are computed with the help of experts and current subway network data. The analysis is done using the Fuzzy Analytic Network Process (FANP) to accommodate the subjectivity of human judgment as being expressed in natural language which entails 'fuzziness' in real-life problems and account for the interdependency between the selected attributes. The output of the model is a criticality based clustering of subway stations. The proposed framework helps authorities prioritize stations for rehabilitation and highlight stations with more criticality for a more robust asset analysis.