Leaf rust of wheat causes considerable losses worldwide. New pathotypes may cause previously resistant varieties to become susceptible. Identification of pathotypes and their relationships provide information for breeding efforts and designing management strategies. Traditional identification of pathotypes is based on responses of differential hosts. At present 50 pathotypes of P. triticina are maintained in the National collection. To determine variability and relationships at the molecular level we conducted analyses with 26 SSR primers, eight of which were polymorphic. Binary (0 or 1) molecular data generated by NTSYS-pc was used to construct a phylogenetic tree. Cluster analysis was done using the unweighted pair group arithmetic means (UPGMA) method in the SAHN program of NTSYS-pc. Pathotype groups and subgroups were determined based on the Jaccard similarity coefficients (JC). Manual observations indicated seven major groups. Among them, two groups each have one pathotype (pathotypes 16 and 17). Jaccard similarity coefficients supported groupings based on pathogenicity data. For example, pathotypes in the race 12 group (12, 12-1, 12-3, 12-4, 12-6, 12-7, 12-8, but excluding 12-2 and 12-5) had similarity coefficients greater than 0.7. Similar observations were recorded for the race 77 group. Maximum similarity was observed between 12-3 and 12-7 (JC value: 0.89) followed by 12-3, 12-7 and 12-6 (JC value: 0.82). Based on the phylogenetic tree and similarity coefficients data, there was substantial diversity among pathotypes. Thus SSR marker data can be used for effective characterization of pathotypes and for making evolutionary inferences.