An intense epidemic of dengue haemorrhagic fever in 1998 prompted the Thai government to investigate the feasibility of focalized vector (Aedes aegypti) control programmes. We tested for correlations of three indices of Aedes larval abundance (housing index, container index and Breteau index) against 38 socio-economic and four climatic variables. Availability of public water wells, existence of transport services and proportion of tin houses were positively associated with larval indices. Private water wells, health education, health insurance coverage, thatched houses and use of firewood for cooking were negatively associated. These probably represent both direct effects on breeding sites (private vs. public wells decrease necessity to store water, and health education may encourage breeding site removal), and more general effects of health-related attitudes, housing quality and remoteness from urban areas. Indices were positively associated with daily minimum temperature, an increase in precipitation from the previous month (reflecting the onset of the rainy season) and daily maximum temperatures of approximately 33-34 degrees C. The associations were used to derive statistical models to predict the rank order of larval indices within the study area (Spearman's correlation coefficients = 0.525-0.554). The study provides a rational basis for identifying possible social interventions, and for prioritizing previously unsurveyed villages for further monitoring and focalized vector control.