It is generally accepted that Aedes aegypti (L.) (Diptera: Culicidae) has a short dispersal capacity, and that displacement can be influenced by the availability of oviposition sites in the surroundings of emergence or release sites. In the present article, we observed the influence of spatial heterogeneity of large containers and human hosts on the cumulative flight direction of Ae. aegypti females during the first gonotrophic cycle, testing the hypothesis that they aggregate in resource-rich areas, i.e. where there are higher concentrations of large containers and/or humans per habitation. We analysed data from pupal surveys and mark-release-recapture experiments (non-blood-fed females were released) carried out in two dengue endemic neighbourhoods of Rio de Janeiro, Brazil: Tubiacanga (a suburb, with a human density of 337 inhabitants/ha) and Favela do Amorim (a slum, with a human density of 901 inhabitants/ha). In both areas, host-seeking females of three different release cohorts showed an overall non-uniform and extensive dispersal from their release point within 1–2 days post-release. At 4–5 days post-release, when many of the released females would be expected to be gravid, in Tubiacanga most mosquitoes were collected in areas with a relatively higher density of containers/premise, independently of the density of residents/house, whereas in Favela do Amorim, almost half of the captured mosquitoes were collected in relatively resource-poorer areas. Although Ae. aegypti dispersal patterns varied between sites, overall the distances travelled from the release point and the cumulative flight directions were correlated with the density of containers and hosts, more markedly in Tubiacanga than in Favela do Amorim.