Using a geographic information system (GIS), the microepidemiology of malaria was studied in an area of 8 square km in Matola, a densely populated periurban are of Maputo, southern Mozambique. Integrating data on the geographic location of the main mosquito breeding sites, the location and characteristics of the house in the study area, and malariometric data, the risk of malaria relation to proximity of the main mosquito breeding sites was described. Inhabitants of five clusters of houses located at different distances from the main mosquito breeding sites, were enrolled in an open cohort and followed for 18 months, from December 1993 to June 1995. Bimonthly cross-sectional surveys, were carried out to estimate the prevalence of malaria in the cohort. Body temperature and consumption of chloroquine were evaluated. Daily visits to all houses included in the cohort allowed the detection off all plasmodium falciparum infected individual’s whit fever or referring a fever episode since the last visit. Cut- off values of parasitemia, above which a fever case at the moment of the visit was considered to be due to clinical malaria, were estimated using data from the cross sectional surveys. Age specific cut-offs were estimated for each age, using a logistic regression model of the risk of fever according to parasite density...