BACKGROUND: Studies of malaria in well-defined cohorts offer important data about the epidemiology of this complex disease, but few have been done in urban African populations. To generate a sampling frame for a longitudinal study of malaria incidence and treatment in Kampala, Uganda, a census, mapping and survey project was conducted. METHODS: All households in a geographically defined area were enumerated and mapped. Probability sampling was used to recruit a representative sample of children and collect baseline descriptive data for future longitudinal studies. RESULTS: 16,172 residents living in 4931 households in a densely-populated community (18,824 persons/km2) were enumerated. A total of 582 households were approached with at least one child less than 10 years of age in order to recruit 601 children living in 322 households. At enrollment, 19% were parasitaemic, 24% were anaemic, 43% used bednets, and 6% used insecticide-treated nets. Low G6PD activity (OR = 0.33, P = 0.009) and bednet use (OR = 0.64, P = 0.045) were associated with a decreased risk of parasitaemia. Increasing age (OR = 0.62 for each year, P < 0.001) and bednet use (OR = 0.58, P = 0.02) were associated with a decreased risk of anaemia CONCLUSION: Detailed surveys of target populations in urban Africa can provide valuable descriptive data and provide a sampling frame for recruitment of representative cohorts for longitudinal studies. Plans to use a multi-disciplinary approach to improve the understanding of the distribution and determinants of malaria incidence and response to therapy in this population are discussed.