OBJECTIVES: The objective of this study was to produce data indicating whether insecticide-treated bednets should replace insecticide house spraying as a malaria control method in South Africa. We report 2 years of preliminary data on malaria incidence comparing areas receiving insecticide-treated bednets and those subjected to house spraying in northern KwaZulu-Natal. DESIGN, SETTING AND SUBJECTS: In order to measure significant reductions in malaria incidence between the two interventions, a geographical information system (GIS) was used to identify and create seven pairs of geographical blocks (areas) in the malaria high-risk areas of Ndumu and Makanis in Ingwavuma magisterial district, KwaZulu-Natal. Individual blocks were then randomly allocated to either insecticide-treated bednets or house spraying with deltamethrin. Malaria cases were either routinely recorded by surveillance agents at home or were reported to the nearest health facility. RESULTS AND CONCLUSIONS: The results show that 2 years' use of insecticide-treated bednets by communities in Ndumu and Makanis, KwaZulu-Natal, significantly reduced the malaria incidence both in 1997 (rate ratio (RR) = 0.879, 95% confidence interval (CI) 0.80-0.95, P = 0.04) and in 1998 (RR = 0.667, CI 0.61-0.72, P = 0.0001). Using a t-test, these significant reductions were further confirmed by an assessment of the rate of change between 1996 and 1998, showing a 16% reduction in malaria incidence in blocks using treated bednets and an increase of 45% in sprayed areas (t = 2.534, P = 0.026 (12 df)). In order to decide whether bednets should replace house spraying in South Africa, we need more data on the efficacy of treated bednets, their long-term acceptability and the cost of the two interventions.