The population densities of Aedes aegypti Linneaus in four towns in Trinidad were studied using standard house-to-house inspections of all water-holding containers to determine whether persistently positive containers and premises existed over a three-month period in the wet season, from May to July 2002. From a total of 1503 houses inspected, 223 were positive with 41 persistently positive over the three month period and classified as 'key premises'. The definition of the term key premises is described and the rationale for its utilization discussed. A total of 24,439 containers was inspected from Santa Margarita (6407), Mt Lambert (5709), St Augustine (5384) and Curepe (6939) of which 1.3% or 334 containers were positive for A. aegypti larvae and pupae. A total of 16,507 immatures of A. aegypti were retrieved from these containers which comprised 17 container types but when these were ranked according to productivity levels, only water drums (average 53.5%), buckets (22.2%), tubs and basins (8.0%), water tanks (5.4%), brick holes (4.2%) and tyres (2.0%) were significant (P < 0.001) producers. The role that key premises play in the introduction and re-infestation of A. aegypti-free communities is described and illustrated. These results suggest that A. aegypti control programmes could be more cost effective and sustainable by concentrating efforts on key premises and key containers to control mosquito densities and Dengue transmission while reducing manpower needs and insecticide use.