Malaria incidence in the province of KwaZulu-Natal is influenced by seasonal climatic variations, drug and insecticide resistance, and population mobility within the region. Current control methods are directed at the malaria vector, the Anopheles arabiensis mosquito, through indoor residual spraying. Control of the dominant malaria parasite, Plasmodium falciparum, is done by treatmenting the symptomatic and asymptomatic parasite carriers with prophylactic medications of. A ceiling of effectiveness with current control efforts have been reached, necessitating the search for supplementary methods. The study area is located in the Ingwavuma District of northern of KwaZulu-Natal which adjoins the Mozambique border to the north, and includes the malaria areas of Ndumu and Makanisdrift. Homestead location coordinates were obtained with Global Positioning System (GPS) receivers and linked to malaria case records through homestead numbers which have been allocated to all homes in the area. The study includes the cases reported during 1993 and 1994 as this was the only data available when the project commenced in 1995. A geographic information system was used to undertake the spatial analysis to test the hypothesis that the malaria vector, Anopheles arabiensis, is localised to certain breeding sites during the winter months in northern KwaZulu-Natal. Identification of these winter 'seed point' breeding sites from which the onset of transmission spreads during the following malaria season, will allow them to be targeted for winter larval control measures. This will contribute to limiting the distribution and lowering the levels of malaria intensity in the region as a whole. The analysis also provided evidence of the maximum likely flight distance of the female mosquito given an adequate host supply in close proximity, thereby identifying those areas requiring additional prevention and control activities. Understanding the local epidemiology of the disease was necessary to determine which monthly malaria cases to include in order to identify the winter breeding sites, due to seasonal variations in the length of the mosquitoes life cycle. Medical geography, as a sub-discipline of geography, combines investigating spatial patterns with the epidemiological principles of medicine and zoology through scientific methods. It is traditionally divided into two approaches, the first being the geography of disease, under which this research falls, and the second being the geography of health care. The integration of the two disciplines allowed the results of the analysis to be presented in maps, graphs and tables in order to describe, interpret, test and explain possible associations between the location of the potential breeding sites and the homesteads at which the malaria cases were reported. The potential breeding sites consisted of the perennial pans, non-perennial pans and dams. Zones were created in the GIS at one kilometre intervals from these sites up to a distance of four kilometres, and the number of cases within each zone determined and corrected for the population at risk per 1000 people for comparative purposes. This spatial analysis was followed by the statistical analysis of the results to verify the findings. The results of the spatial and statistical analysis indicated that the perennial pans were used as the winter 'seed point' breeding sites, and that the maximum likely flight distance of the female mosquito, given an adequate host supply in close proximity, is 4 kilometres. The results will be made available to the local malaria research and control community who will assess the feasibility of implementing supplementary control measures. / Thesis (M.Sc.)-University of Natal, Durban, 2001.