To increase the effectiveness of ongoing anti-dengue control measures, we conducted a case-control study in a favela in Fortaleza (north-east Brazil) during an outbreak of dengue fever. Cases were defined according to the national guidelines for dengue control as well as based on the detection of IgM-dengue antibodies, and 34 cases and 34 controls were investigated. Significant risk factors were: living in a street perpendicular to the beach on which the favela is situated (P < 0.0001), an interval > 30 days since the last visit of the vector control agent (P=0.001), receptacles in the garden or courtyard (P=0.001), plants with temporary water pools on the property, gutter to collect rainwater, uncovered water storage container (all P=0.02), and no waste collection (P=0.03). Socio-economic variables were not associated with dengue fever. The probable starting point of the epidemic was an uncovered water tank on the roof of the house adjacent to the index case. From there, the outbreak spread uphill parallel to the prevailing direction of the wind. Thus, the chronological and spatial evolution of the epidemic could have been forecast after the first cases had occurred. This example of investigative epidemiology in an operational setting shows that targeted intervention leading to increased effectiveness of control measures is possible.