The objective of this work was to define criteria that could be applied to achieve faster, more economical, and accurate assessment of vector populations for control of dengue viruses. During 1989-1990, 1,349 premises were surveyed in Townsville, Charters Towers and Mingela/Ravenswood, Queensland, Australia. In each locality, 1.9-8.4% of premises contained three or more containers with Aedes aegypti immature forms and were designated as key premises. Comparison of surveys in Townsville from 1989 to 1990 indicated that positive premises (i.e., those with at least one container with Ae. aegypti present) were 3.22 times more likely to remain positive than negative houses to become positive the following year. The Ae. aegypti population in Townsville was seen to be totally associated with garden receptacles, discarded household items, and trash but one well and one rainwater tank were responsible for 28% of all immature forms recorded in the 1,349 premises inspected. These breeding sites of high productivity were designated as key containers. At Charters Towers, Mingela, and Ravenswood, rainwater tanks were seen as the most important key container because although they constituted 13-29% of positive containers, they supported 60-63% of the immature forms. This study demonstrates that there is a certain degree of stability with regard to positive premises and that some of these, or some container types, contribute disproportionately to the Ae. aegypti population. Control programs could be made more efficient if efforts were concentrated on these sites of key vector productivity.