Malaria in Sri Lanka is unstable and epidemic, with large spatial and temporal differences in transmission dynamics. The disease is of great public health significance and identification of underlying risk factors is important in order to use the limited resources in a cost-effective way. The International Water Management Institute (IWMI) recently launched a project of GIS-based malaria risk mapping in Sri Lanka, to investigate whether this tool could be used for epidemic forecasting and for the planning of malaria control activities. This paper presents results for the Uda Walawe region in southern Sri Lanka, an irrigated agricultural area where malaria cases were mapped at the smallest administrative level for each month over a 10-year period. Malaria incidence rates were related to land- and water-use patterns, socio-economic features, and data on malaria control interventions in a multivariate analysis. Areas of high malaria risk were characterized by: (i) higher than average rainfall, (ii) greater forest coverage; (iii) slash and burn cultivation as a predominant agricultural activity; (iv) presence of many abandoned irrigation reservoirs; and (v) poor socio-economic status. Irrigated rice cultivation areas had a lower risk of malaria than non-irrigated areas. This difference could be due to socio-economic factors related to irrigation development and/or transmission dynamics related to vector density or species composition. Our findings call for malaria control strategies that are readily adapted to different ecological and epidemiological settings. Malaria risk maps are a convenient tool for discussing targeted and cost-effective interventions with disease control personnel.