The incidence of malaria in Solomon Islands has been decreasing since 1992. The control program used a combination of methods including DDT residual house spraying and insecticide-treated mosquito nets. To determine how much each method contributed to malaria control, data were analyzed on monthly incidence and on control activities for 41 of 110 malaria zones over the same time period (January 1993 to August 1999). After correction for endogeneity, then spraying, insecticide treatment of nets, and education about malaria are all independently associated with reduction in incident cases of malaria or fever, while larviciding with temephos is not. The evidence suggests that although impregnated bed nets cannot entirely replace DDT spraying without substantial increase in incidence, their use permits reduced DDT spraying. The paper shows that non-experimental data can be used to infer causal links in epidemiology, provided that instrumental variables are available to correct for endogeneity.