Dengue has become the fastest-growing mosquito-borne disease in Sri Lanka and the control of the vectors, Aedes aegypti Linnaeus and Ae. albopictus Skuse, is the most effective way of controlling the disease. A detailed study on vector prevalence has not been recorded from Sri Lanka. Therefore, the present study was undertaken to study the prevalence of both vectors in four semi-urban study sites in two of the most affected districts, namely Kandy (wet zone) and Kurunegala (intermediate zone), by conducting egg surveys (using ovitraps) and larval surveys from June 2007 to May 2008. A total of 82,524 eggs and 2658 larvae of Ae. aegypti and Ae. albopictus were collected. A total of 3699 potential breeding habitats were examined. Ovitrap and larval indices (house, container and Breteau) showed that all four areas are at epidemic risk, especially due to a high abundance of Ae. albopictus. The highest numbers for both the species were from the Kandy sites where dense vegetation, high rainfall and low temperature prevailed. The results showed a high mortality rate during the egg-to-larva transition, suggesting that conducting an egg survey alone would overestimate the vector abundance and the disease risk. For Ae. albopictus, the monthly mean number of eggs showed positive relationships with relative humidity in both districts and with rainfall in the Kandy District. The number of dengue cases in the area had no positive relationships with the abundance of eggs or larval density indices. Discarded receptacles were the most preferred breeding habitat for these dengue vectors. Since the attractiveness (inferred by the calculated risk factors) of most of the breeding habitats was very high, elimination of these breeding sites is essential for the success of dengue control programmes.