To develop a specific, targeted intervention strategy for reducing preterm birth through use of geographic analysis. We utilized Hamilton County, Ohio vital records and Census data from 2003 to 2006. Spatial scanning statistics allowed determination of the prevalence of preterm birth for any geographical point. Attributable risk calculations demonstrated heterogeneity of risk factors within areas of high or low preterm birth prevalence. Three geographically separate areas with high preterm birth proportions (>16%) had differing primary risk factors for preterm birth, including short interpregnancy interval, previous preterm birth, and low prepregnancy weight, despite similarities in demographics and physical location. Primary risk factors also differed when comparing areas with high and low preterm birth proportions, with diabetes and smoking having primary associations in the lower risk areas. Each local region of high preterm birth proportion as well as those with average or low proportion displayed distinct hierarchies of attributable risk. The heterogeneous distribution of preterm birth proportion within an urban county is complex and requires location specific analysis to develop targeted interventions.