In this article we discuss the methodological issues associated with the creation of a surveillance system for endemic diseases in urban areas based on analysis of populations at risk and on spatially referenced epidemiological indicators. We comment on the system's basic requirements, selection criteria for socioeconomic variables, and methodological steps to combine these variables so as to construct a census-based deprivation index. We also present the ways we solved some operational problems related to generation of digitized census tracts maps and linkage of morbidity data from different sources. This approach, spatial organization into account in surveillance of endemic diseases, exemplified here by tuberculosis and leprosy, allows for the interaction of several official data sets from census and health services in order to geographically discriminate inner-city risk strata. Criteria for constructing these risk strata were considered a useful tool for health planning and management activities for the control of endemic diseases in cities.