Opportunistic feeding behaviour and partial exophily make An. arabiensis much more difficult to control by indoor residual spraying than any other vector in the Afro-tropical region. The persistent malaria outbreaks in Zimbabwe despite decades of indoor house spraying prompted this investigation into the role of An. arabiensis in malaria transmission and assessment of the possible impact of this control measure. The study was conducted in the malaria endemic districts of Binga and Gokwe. An. gambiae complex mosquitoes were collected from artificial outdoor resting sites, and from human dwellings by i) daytime hut searches, ii) pyrethrum spray catches and iii) exit window traps. Mosquito components were processed to enable: i) the distinction of An. arabiensis from An. quadriannulatus and An. merus on the basis of the pale band at the junction of the hind leg 3/4 tarsomeres; ii) species identification and scoring of inversion polymorphism on the basis of the X chromosome and autosomes respectively; iii) the determination of blood meal sources using the Ouchterlony precipitin test; and iv) identification of An. gambiae s.l. using polymerase chain reaction (PCR) and enzyme electrophoresis techniques. Entomological assessment of residual spraying included determining: the vector resting densities indoors and outdoors, bioassay and insecticides susceptibility tests. Data were also collected on hut profiles, knowledge-attitudes-practices surveys, and household malaria prevalence . surveys. An. arabiensis and An. quadriannulatus were found in sympatry in Binga and Gokwe, and in addition, An. merus was found in Gokwe. Most species identifications were made using PCR; which was found to have 7.5% and 41.6% levels of error for An. arabiensis and An. quadriannulatus respectively, using the cytogenetic technique as benchmark. The pale band technique yielded > 80% correct identification for An. arabiensis but the extent of overlap in the pale band lengths between An. arabiensis and An. quadriannulatus renders the method unsuitable for distinguishing these two species. Inversions 2Rb and 3Ra were found floating in An. arabiensis, with 60% frequency in the former. The Wright's F statistic value of -0.0416 indicated an excess of heterozygotes, and a state of panmixis in the vector population. No significant differences were observed between 2Rb karyotypes in host choice. Human blood indices among indoor (0.82), exit trap (0.98) and outdoor resting (0.30) specimens suggested exophilic behaviour. This was corroborated by the high fed:gravid ratios of 6.8: 1 and 11.6: 1 in sprayed and non-sprayed dwellings respectively. This was worsened by a high feeder-survivor index (FSI) of 93 % among exit trap specimens. The susceptibility to deltamethrin coupled with residual efficacy nine weeks post-spray indicated the suitability of the insecticide. Rural dwellings were suitably built for spraying but had no mosquito proofing. Personal protective measures are hardly known; sleeping outdoors occurs in Siabuwa. While An. arabiensis bites humans indoors the partial exophily it exhibits is a threat to indoor residual insecticide spraying. An integrated malaria control approach is recommended. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1996.