Transmission characteristics of malaria were studied in Matola, a coastal suburb of Maputo, the capital City, in southern Mozambique, from November 1994 to April 1996. The local climate alternates between cool dry season (May-October) and hot rainy season (November-April) with mean annual rainfall 650-850 mm. Saltmarsh and freshwater pools provide mosquito breeding sites in Matola. Malaria prevalence reached approximately 60% among people living nearest to the main breeding sites of the vectors. Plasmodium falciparum caused 97% of malaria cases, others being P. malariae and P. ovale. Potential malaria vector mosquitoes (Diptera: Culicidae) collected at Matola during daytime indoor-resting (n = 1021) and on human bait at night (n = 5893) comprised 12% Anopheles coustani Laveran (93% biting outdoors), 46% An. funestus Giles (68% biting indoors) and 42% An. gambiae Giles sensu lato (60% biting outdoors). All 215 specimens of An. gambiae s.l. identified genetically were An. arabiensis Patton. Anopheles funestus populations remained stable throughout the year, whereas densities of the An. gambiae complex fluctuated considerably, with An. arabiensis peaking during the rainy season. No concomitant rise in malaria incidence was observed. Human landing indices of An. funestus and An. arabiensis averaged 1.8 and 3.8 per man-night, respectively. Overall Plasmodium sporozoite rates were 2.42+/-1.24% in 2181 An. funestus and 1.11+/-1.25% in 1689 An. arabiensis dissected and examined microscopically. Mean daily survival rates were 0.79 for both vector species. Estimated infective bites/person/year were 15 An. funestus and 12 An. arabiensis. Biting rates were greatest at 2100-24.00 hours for An. funestus (68% endophagic) and 21.00-03.00 hours for An. arabiensis (40% endophagic). The entomological inoculation rate (EIR) declined sharply over very short distances (50% per 90m) away from breeding-sites of the vectors. Consequently, P. falciparum prevalence among Matola residents was halved 350 m within the town. Implications for the protective effectiveness of a 'cordon sanitaire' by residual house-spraying and/or the use of insecticide-treated bednets are discussed.